diff options
Diffstat (limited to '')
| -rw-r--r-- | registry/Makefile.am | 8 | ||||
| -rw-r--r-- | registry/README.md | 203 | ||||
| -rw-r--r-- | registry/registry.c | 435 | ||||
| -rw-r--r-- | registry/registry.h | 83 | ||||
| -rw-r--r-- | registry/registry_db.c | 346 | ||||
| -rw-r--r-- | registry/registry_init.c | 147 | ||||
| -rw-r--r-- | registry/registry_internals.c | 332 | ||||
| -rw-r--r-- | registry/registry_internals.h | 89 | ||||
| -rw-r--r-- | registry/registry_log.c | 136 | ||||
| -rw-r--r-- | registry/registry_machine.c | 104 | ||||
| -rw-r--r-- | registry/registry_machine.h | 43 | ||||
| -rw-r--r-- | registry/registry_person.c | 267 | ||||
| -rw-r--r-- | registry/registry_person.h | 62 | ||||
| -rw-r--r-- | registry/registry_url.c | 88 | ||||
| -rw-r--r-- | registry/registry_url.h | 35 |
15 files changed, 2378 insertions, 0 deletions
diff --git a/registry/Makefile.am b/registry/Makefile.am new file mode 100644 index 0000000..161784b --- /dev/null +++ b/registry/Makefile.am @@ -0,0 +1,8 @@ +# SPDX-License-Identifier: GPL-3.0-or-later + +AUTOMAKE_OPTIONS = subdir-objects +MAINTAINERCLEANFILES = $(srcdir)/Makefile.in + +dist_noinst_DATA = \ + README.md \ + $(NULL) diff --git a/registry/README.md b/registry/README.md new file mode 100644 index 0000000..968292c --- /dev/null +++ b/registry/README.md @@ -0,0 +1,203 @@ +<!-- +title: "Registry" +description: "Netdata utilizes a central registry of machines/person GUIDs, URLs, and opt-in account information to provide unified cross-server dashboards." +custom_edit_url: https://github.com/netdata/netdata/edit/master/registry/README.md +--> + +# Registry + +Netdata provides distributed monitoring. + +Traditional monitoring solutions centralize all the data to provide unified dashboards across all servers. Before +Netdata, this was the standard practice. However it has a few issues: + +1. due to the resources required, the number of metrics collected is limited. +2. for the same reason, the data collection frequency is not that high, at best it will be once every 10 or 15 seconds, + at worst every 5 or 10 mins. +3. the central monitoring solution needs dedicated resources, thus becoming "another bottleneck" in the whole + ecosystem. It also requires maintenance, administration, etc. +4. most centralized monitoring solutions are usually only good for presenting _statistics of past performance_ (i.e. + cannot be used for real-time performance troubleshooting). + +Netdata follows a different approach: + +1. data collection happens per second +2. thousands of metrics per server are collected +3. data do not leave the server where they are collected +4. Netdata servers do not talk to each other +5. your browser connects all the Netdata servers + +Using Netdata, your monitoring infrastructure is embedded on each server, limiting significantly the need of additional +resources. Netdata is blazingly fast, very resource efficient and utilizes server resources that already exist and are +spare (on each server). This allows **scaling out** the monitoring infrastructure. + +However, the Netdata approach introduces a few new issues that need to be addressed, one being **the list of Netdata we +have installed**, i.e. the URLs our Netdata servers are listening. + +To solve this, Netdata utilizes a **central registry**. This registry, together with certain browser features, allow +Netdata to provide unified cross-server dashboards. For example, when you jump from server to server using the node +menu, several session settings (like the currently viewed charts, the current zoom and pan operations on the charts, +etc.) are propagated to the new server, so that the new dashboard will come with exactly the same view. + +## What data does the registry store? + +The registry keeps track of 4 entities: + +1. **machines**: i.e. the Netdata installations (a random GUID generated by each Netdata the first time it starts; we + call this **machine_guid**) + + For each Netdata installation (each `machine_guid`) the registry keeps track of the different URLs it has accessed. + +2. **persons**: i.e. the web browsers accessing the Netdata installations (a random GUID generated by the registry the + first time it sees a new web browser; we call this **person_guid**) + + For each person, the registry keeps track of the Netdata installations it has accessed and their URLs. + +3. **URLs** of Netdata installations (as seen by the web browsers) + + For each URL, the registry keeps the URL and nothing more. Each URL is linked to _persons_ and _machines_. The only + way to find a URL is to know its **machine_guid** or have a **person_guid** it is linked to it. + +4. **accounts**: i.e. the information used to sign-in via one of the available sign-in methods. Depending on the + method, this may include an email, or an email and a profile picture or avatar. + +For _persons_/_accounts_ and _machines_, the registry keeps links to _URLs_, each link with 2 timestamps (first time +seen, last time seen) and a counter (number of times it has been seen). *machines_, _persons_ and timestamps are stored +in the Netdata registry regardless of whether you sign in or not. + +## Who talks to the registry? + +Your web browser **only**! If sending this information is against your policies, you can [run your own +registry](#run-your-own-registry) + +Your Netdata servers do not talk to the registry. This is a UML diagram of its operation: + + + +## Which is the default registry? + +`https://registry.my-netdata.io`, which is currently served by `https://london.my-netdata.io`. This registry listens to +both HTTP and HTTPS requests but the default is HTTPS. + +### Can this registry handle the global load of Netdata installations? + +Yeap! The registry can handle 50.000 - 100.000 requests **per second per core** (depending on the type of CPU, the +computer's memory bandwidth, etc). 50.000 is on J1900 (celeron 2Ghz). + +We believe, it can do it... + +## Run your own registry + +**Every Netdata can be a registry**. Just pick one and configure it. + +**To turn any Netdata into a registry**, edit `/etc/netdata/netdata.conf` and set: + +```conf +[registry] + enabled = yes + registry to announce = http://your.registry:19999 +``` + +Restart your Netdata to activate it. + +Then, you need to tell **all your other Netdata servers to advertise your registry**, instead of the default. To do +this, on each of your Netdata servers, edit `/etc/netdata/netdata.conf` and set: + +```conf +[registry] + enabled = no + registry to announce = http://your.registry:19999 +``` + +Note that we have not enabled the registry on the other servers. Only one Netdata (the registry) needs +`[registry].enabled = yes`. + +This is it. You have your registry now. + +You may also want to give your server different names under the node menu (i.e. to have them sorted / grouped). You can +change its registry name, by setting on each Netdata server: + +```conf +[registry] + registry hostname = Group1 - Master DB +``` + +So this server will appear in the node menu as `Group1 - Master DB`. The max name length is 50 characters. + +### Limiting access to the registry + +Netdata v1.9+ support limiting access to the registry from given IPs, like this: + +```conf +[registry] + allow from = * +``` + +`allow from` settings are [Netdata simple patterns](/libnetdata/simple_pattern/README.md): string matches that use `*` +as wildcard (any number of times) and a `!` prefix for a negative match. So: `allow from = !10.1.2.3 10.*` will allow +all IPs in `10.*` except `10.1.2.3`. The order is important: left to right, the first positive or negative match is +used. + +Keep in mind that connections to Netdata API ports are filtered by `[web].allow connections from`. So, IPs allowed by +`[registry].allow from` should also be allowed by `[web].allow connection from`. + +The patterns can be matches over IP addresses or FQDN of the host. In order to check the FQDN of the connection without +opening the Netdata agent to DNS-spoofing, a reverse-dns record must be setup for the connecting host. At connection +time the reverse-dns of the peer IP address is resolved, and a forward DNS resolution is made to validate the IP address +against the name-pattern. + +Please note that this process can be expensive on a machine that is serving many connections. The behaviour of the +pattern matching can be controlled with the following setting: + +```conf +[registry] + allow by dns = heuristic +``` + +The settings are: +- `yes` allows the pattern to match DNS names. +- `no` disables DNS matching for the patterns (they only match IP addresses). +- `heuristic` will estimate if the patterns should match FQDNs by the presence or absence of `:`s or alpha-characters. + +### Where is the registry database stored? + +`/var/lib/netdata/registry/*.db` + +There can be up to 2 files: + +- `registry-log.db`, the transaction log + + all incoming requests that affect the registry are saved in this file in real-time. + +- `registry.db`, the database + + every `[registry].registry save db every new entries` entries in `registry-log.db`, Netdata will save its database + to `registry.db` and empty `registry-log.db`. + +Both files are machine readable text files. + +## The future + +The registry opens a whole world of new possibilities for Netdata. Check here what we think: +<https://github.com/netdata/netdata/issues/416> + +## Troubleshooting the registry + +The registry URL should be set to the URL of a Netdata dashboard. This server has to have `[registry].enabled = yes`. +So, accessing the registry URL directly with your web browser, should present the dashboard of the Netdata operating the +registry. + +To use the registry, your web browser needs to support **third party cookies**, since the cookies are set by the +registry while you are browsing the dashboard of another Netdata server. The registry, the first time it sees a new web +browser it tries to figure if the web browser has cookies enabled or not. It does this by setting a cookie and +redirecting the browser back to itself hoping that it will receive the cookie. If it does not receive the cookie, the +registry will keep redirecting your web browser back to itself, which after a few redirects will fail with an error like +this: + +```conf +ERROR 409: Cannot ACCESS netdata registry: https://registry.my-netdata.io responded with: {"status":"redirect","registry":"https://registry.my-netdata.io"} +``` + +This error is printed on your web browser console (press F12 on your browser to see it). + +[](<>) diff --git a/registry/registry.c b/registry/registry.c new file mode 100644 index 0000000..b14f4ee --- /dev/null +++ b/registry/registry.c @@ -0,0 +1,435 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +#define REGISTRY_STATUS_OK "ok" +#define REGISTRY_STATUS_FAILED "failed" +#define REGISTRY_STATUS_DISABLED "disabled" + +// ---------------------------------------------------------------------------- +// REGISTRY concurrency locking + +static inline void registry_lock(void) { + netdata_mutex_lock(®istry.lock); +} + +static inline void registry_unlock(void) { + netdata_mutex_unlock(®istry.lock); +} + + +// ---------------------------------------------------------------------------- +// COOKIES + +static void registry_set_cookie(struct web_client *w, const char *guid) { + char edate[100]; + time_t et = now_realtime_sec() + registry.persons_expiration; + struct tm etmbuf, *etm = gmtime_r(&et, &etmbuf); + strftime(edate, sizeof(edate), "%a, %d %b %Y %H:%M:%S %Z", etm); + + snprintfz(w->cookie1, NETDATA_WEB_REQUEST_COOKIE_SIZE, NETDATA_REGISTRY_COOKIE_NAME "=%s; Expires=%s", guid, edate); + + if(registry.registry_domain && registry.registry_domain[0]) + snprintfz(w->cookie2, NETDATA_WEB_REQUEST_COOKIE_SIZE, NETDATA_REGISTRY_COOKIE_NAME "=%s; Domain=%s; Expires=%s", guid, registry.registry_domain, edate); +} + +static inline void registry_set_person_cookie(struct web_client *w, REGISTRY_PERSON *p) { + registry_set_cookie(w, p->guid); +} + + +// ---------------------------------------------------------------------------- +// JSON GENERATION + +static inline void registry_json_header(RRDHOST *host, struct web_client *w, const char *action, const char *status) { + buffer_flush(w->response.data); + w->response.data->contenttype = CT_APPLICATION_JSON; + buffer_sprintf(w->response.data, "{\n\t\"action\": \"%s\",\n\t\"status\": \"%s\",\n\t\"hostname\": \"%s\",\n\t\"machine_guid\": \"%s\"", + action, status, host->registry_hostname, host->machine_guid); +} + +static inline void registry_json_footer(struct web_client *w) { + buffer_strcat(w->response.data, "\n}\n"); +} + +static inline int registry_json_disabled(RRDHOST *host, struct web_client *w, const char *action) { + registry_json_header(host, w, action, REGISTRY_STATUS_DISABLED); + + buffer_sprintf(w->response.data, ",\n\t\"registry\": \"%s\"", + registry.registry_to_announce); + + registry_json_footer(w); + return 200; +} + + +// ---------------------------------------------------------------------------- +// CALLBACKS FOR WALKING THROUGH REGISTRY OBJECTS + +// structure used be the callbacks below +struct registry_json_walk_person_urls_callback { + REGISTRY_PERSON *p; + REGISTRY_MACHINE *m; + struct web_client *w; + int count; +}; + +// callback for rendering PERSON_URLs +static int registry_json_person_url_callback(void *entry, void *data) { + REGISTRY_PERSON_URL *pu = (REGISTRY_PERSON_URL *)entry; + struct registry_json_walk_person_urls_callback *c = (struct registry_json_walk_person_urls_callback *)data; + struct web_client *w = c->w; + + if (!strcmp(pu->url->url,"***")) return 0; + + if(unlikely(c->count++)) + buffer_strcat(w->response.data, ","); + + buffer_sprintf(w->response.data, "\n\t\t[ \"%s\", \"%s\", %u000, %u, \"%s\" ]", + pu->machine->guid, pu->url->url, pu->last_t, pu->usages, pu->machine_name); + + return 0; +} + +// callback for rendering MACHINE_URLs +static int registry_json_machine_url_callback(void *entry, void *data) { + REGISTRY_MACHINE_URL *mu = (REGISTRY_MACHINE_URL *)entry; + struct registry_json_walk_person_urls_callback *c = (struct registry_json_walk_person_urls_callback *)data; + struct web_client *w = c->w; + REGISTRY_MACHINE *m = c->m; + + if (!strcmp(mu->url->url,"***")) return 1; + + if(unlikely(c->count++)) + buffer_strcat(w->response.data, ","); + + buffer_sprintf(w->response.data, "\n\t\t[ \"%s\", \"%s\", %u000, %u ]", + m->guid, mu->url->url, mu->last_t, mu->usages); + + return 1; +} + +// ---------------------------------------------------------------------------- + +// structure used be the callbacks below +struct registry_person_url_callback_verify_machine_exists_data { + REGISTRY_MACHINE *m; + int count; +}; + +static inline int registry_person_url_callback_verify_machine_exists(void *entry, void *data) { + struct registry_person_url_callback_verify_machine_exists_data *d = (struct registry_person_url_callback_verify_machine_exists_data *)data; + REGISTRY_PERSON_URL *pu = (REGISTRY_PERSON_URL *)entry; + REGISTRY_MACHINE *m = d->m; + + if(pu->machine == m) + d->count++; + + return 0; +} + +// ---------------------------------------------------------------------------- +// dynamic update of the configuration +// The registry does not seem to be designed to support this and I cannot see any concurrency protection +// that could make this safe, so try to be as atomic as possible. + +void registry_update_cloud_base_url() +{ + // This is guaranteed to be set early in main via post_conf_load() + registry.cloud_base_url = appconfig_get(&cloud_config, CONFIG_SECTION_GLOBAL, "cloud base url", NULL); + if (registry.cloud_base_url == NULL) + fatal("Do not move the cloud base url out of post_conf_load!!"); + + setenv("NETDATA_REGISTRY_CLOUD_BASE_URL", registry.cloud_base_url, 1); +} +// ---------------------------------------------------------------------------- +// public HELLO request + +int registry_request_hello_json(RRDHOST *host, struct web_client *w) { + registry_json_header(host, w, "hello", REGISTRY_STATUS_OK); + + buffer_sprintf(w->response.data, + ",\n\t\"registry\": \"%s\",\n\t\"cloud_base_url\": \"%s\",\n\t\"anonymous_statistics\": %s", + registry.registry_to_announce, + registry.cloud_base_url, netdata_anonymous_statistics_enabled?"true":"false"); + + registry_json_footer(w); + return 200; +} + +// ---------------------------------------------------------------------------- +//public ACCESS request + +#define REGISTRY_VERIFY_COOKIES_GUID "give-me-back-this-cookie-now--please" + +// the main method for registering an access +int registry_request_access_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *name, time_t when) { + if(unlikely(!registry.enabled)) + return registry_json_disabled(host, w, "access"); + + // ------------------------------------------------------------------------ + // verify the browser supports cookies + + if(registry.verify_cookies_redirects > 0 && !person_guid[0]) { + buffer_flush(w->response.data); + registry_set_cookie(w, REGISTRY_VERIFY_COOKIES_GUID); + w->response.data->contenttype = CT_APPLICATION_JSON; + buffer_sprintf(w->response.data, "{ \"status\": \"redirect\", \"registry\": \"%s\" }", registry.registry_to_announce); + return 200; + } + + if(unlikely(person_guid[0] && !strcmp(person_guid, REGISTRY_VERIFY_COOKIES_GUID))) + person_guid[0] = '\0'; + + // ------------------------------------------------------------------------ + + registry_lock(); + + REGISTRY_PERSON *p = registry_request_access(person_guid, machine_guid, url, name, when); + if(!p) { + registry_json_header(host, w, "access", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 412; + } + + // set the cookie + registry_set_person_cookie(w, p); + + // generate the response + registry_json_header(host, w, "access", REGISTRY_STATUS_OK); + + buffer_sprintf(w->response.data, ",\n\t\"person_guid\": \"%s\",\n\t\"urls\": [", p->guid); + struct registry_json_walk_person_urls_callback c = { p, NULL, w, 0 }; + avl_traverse(&p->person_urls, registry_json_person_url_callback, &c); + buffer_strcat(w->response.data, "\n\t]\n"); + + registry_json_footer(w); + registry_unlock(); + return 200; +} + +// ---------------------------------------------------------------------------- +// public DELETE request + +// the main method for deleting a URL from a person +int registry_request_delete_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *delete_url, time_t when) { + if(!registry.enabled) + return registry_json_disabled(host, w, "delete"); + + registry_lock(); + + REGISTRY_PERSON *p = registry_request_delete(person_guid, machine_guid, url, delete_url, when); + if(!p) { + registry_json_header(host, w, "delete", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 412; + } + + // generate the response + registry_json_header(host, w, "delete", REGISTRY_STATUS_OK); + registry_json_footer(w); + registry_unlock(); + return 200; +} + +// ---------------------------------------------------------------------------- +// public SEARCH request + +// the main method for searching the URLs of a netdata +int registry_request_search_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *request_machine, time_t when) { + if(!registry.enabled) + return registry_json_disabled(host, w, "search"); + + registry_lock(); + + REGISTRY_MACHINE *m = registry_request_machine(person_guid, machine_guid, url, request_machine, when); + if(!m) { + registry_json_header(host, w, "search", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 404; + } + + registry_json_header(host, w, "search", REGISTRY_STATUS_OK); + + buffer_strcat(w->response.data, ",\n\t\"urls\": ["); + struct registry_json_walk_person_urls_callback c = { NULL, m, w, 0 }; + dictionary_get_all(m->machine_urls, registry_json_machine_url_callback, &c); + buffer_strcat(w->response.data, "\n\t]\n"); + + registry_json_footer(w); + registry_unlock(); + return 200; +} + +// ---------------------------------------------------------------------------- +// SWITCH REQUEST + +// the main method for switching user identity +int registry_request_switch_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *new_person_guid, time_t when) { + if(!registry.enabled) + return registry_json_disabled(host, w, "switch"); + + (void)url; + (void)when; + + registry_lock(); + + REGISTRY_PERSON *op = registry_person_find(person_guid); + if(!op) { + registry_json_header(host, w, "switch", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 430; + } + + REGISTRY_PERSON *np = registry_person_find(new_person_guid); + if(!np) { + registry_json_header(host, w, "switch", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 431; + } + + REGISTRY_MACHINE *m = registry_machine_find(machine_guid); + if(!m) { + registry_json_header(host, w, "switch", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 432; + } + + struct registry_person_url_callback_verify_machine_exists_data data = { m, 0 }; + + // verify the old person has access to this machine + avl_traverse(&op->person_urls, registry_person_url_callback_verify_machine_exists, &data); + if(!data.count) { + registry_json_header(host, w, "switch", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 433; + } + + // verify the new person has access to this machine + data.count = 0; + avl_traverse(&np->person_urls, registry_person_url_callback_verify_machine_exists, &data); + if(!data.count) { + registry_json_header(host, w, "switch", REGISTRY_STATUS_FAILED); + registry_json_footer(w); + registry_unlock(); + return 434; + } + + // set the cookie of the new person + // the user just switched identity + registry_set_person_cookie(w, np); + + // generate the response + registry_json_header(host, w, "switch", REGISTRY_STATUS_OK); + buffer_sprintf(w->response.data, ",\n\t\"person_guid\": \"%s\"", np->guid); + registry_json_footer(w); + + registry_unlock(); + return 200; +} + +// ---------------------------------------------------------------------------- +// STATISTICS + +void registry_statistics(void) { + if(!registry.enabled) return; + + static RRDSET *sts = NULL, *stc = NULL, *stm = NULL; + + if(unlikely(!sts)) { + sts = rrdset_create_localhost( + "netdata" + , "registry_sessions" + , NULL + , "registry" + , NULL + , "NetData Registry Sessions" + , "sessions" + , "registry" + , "stats" + , 131000 + , localhost->rrd_update_every + , RRDSET_TYPE_LINE + ); + + rrddim_add(sts, "sessions", NULL, 1, 1, RRD_ALGORITHM_ABSOLUTE); + } + else rrdset_next(sts); + + rrddim_set(sts, "sessions", registry.usages_count); + rrdset_done(sts); + + // ------------------------------------------------------------------------ + + if(unlikely(!stc)) { + stc = rrdset_create_localhost( + "netdata" + , "registry_entries" + , NULL + , "registry" + , NULL + , "NetData Registry Entries" + , "entries" + , "registry" + , "stats" + , 131100 + , localhost->rrd_update_every + , RRDSET_TYPE_LINE + ); + + rrddim_add(stc, "persons", NULL, 1, 1, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stc, "machines", NULL, 1, 1, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stc, "urls", NULL, 1, 1, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stc, "persons_urls", NULL, 1, 1, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stc, "machines_urls", NULL, 1, 1, RRD_ALGORITHM_ABSOLUTE); + } + else rrdset_next(stc); + + rrddim_set(stc, "persons", registry.persons_count); + rrddim_set(stc, "machines", registry.machines_count); + rrddim_set(stc, "urls", registry.urls_count); + rrddim_set(stc, "persons_urls", registry.persons_urls_count); + rrddim_set(stc, "machines_urls", registry.machines_urls_count); + rrdset_done(stc); + + // ------------------------------------------------------------------------ + + if(unlikely(!stm)) { + stm = rrdset_create_localhost( + "netdata" + , "registry_mem" + , NULL + , "registry" + , NULL + , "NetData Registry Memory" + , "KiB" + , "registry" + , "stats" + , 131300 + , localhost->rrd_update_every + , RRDSET_TYPE_STACKED + ); + + rrddim_add(stm, "persons", NULL, 1, 1024, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stm, "machines", NULL, 1, 1024, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stm, "urls", NULL, 1, 1024, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stm, "persons_urls", NULL, 1, 1024, RRD_ALGORITHM_ABSOLUTE); + rrddim_add(stm, "machines_urls", NULL, 1, 1024, RRD_ALGORITHM_ABSOLUTE); + } + else rrdset_next(stm); + + rrddim_set(stm, "persons", registry.persons_memory + registry.persons_count * sizeof(NAME_VALUE) + sizeof(DICTIONARY)); + rrddim_set(stm, "machines", registry.machines_memory + registry.machines_count * sizeof(NAME_VALUE) + sizeof(DICTIONARY)); + rrddim_set(stm, "urls", registry.urls_memory); + rrddim_set(stm, "persons_urls", registry.persons_urls_memory); + rrddim_set(stm, "machines_urls", registry.machines_urls_memory + registry.machines_count * sizeof(DICTIONARY) + registry.machines_urls_count * sizeof(NAME_VALUE)); + rrdset_done(stm); +} diff --git a/registry/registry.h b/registry/registry.h new file mode 100644 index 0000000..44095c2 --- /dev/null +++ b/registry/registry.h @@ -0,0 +1,83 @@ +// SPDX-License-Identifier: GPL-3.0-or-later +/* + * netdata registry + * + * this header file describes the public interface + * to the netdata registry + * + * only these high level functions are exposed + * + */ + +// ---------------------------------------------------------------------------- +// TODO +// +// 1. the default tracking cookie expires in 1 year, but the persons are not +// removed from the db - this means the database only grows - ideally the +// database should be cleaned in registry_db_save() for both on-disk and +// on-memory entries. +// +// Cleanup: +// i. Find all the PERSONs that have expired cookie +// ii. For each of their PERSON_URLs: +// - decrement the linked MACHINE links +// - if the linked MACHINE has no other links, remove the linked MACHINE too +// - remove the PERSON_URL +// +// 2. add protection to prevent abusing the registry by flooding it with +// requests to fill the memory and crash it. +// +// Possible protections: +// - limit the number of URLs per person +// - limit the number of URLs per machine +// - limit the number of persons +// - limit the number of machines +// - [DONE] limit the size of URLs +// - [DONE] limit the size of PERSON_URL names +// - limit the number of requests that add data to the registry, +// per client IP per hour +// +// 3. lower memory requirements +// +// - embed avl structures directly into registry objects, instead of DICTIONARY +// [DONE for PERSON_URLs, PENDING for MACHINE_URLs] +// - store GUIDs in memory as UUID instead of char * +// - do not track persons using the demo machines only +// (i.e. start tracking them only when they access a non-demo machine) +// - [DONE] do not track custom dashboards by default + +#ifndef NETDATA_REGISTRY_H +#define NETDATA_REGISTRY_H 1 + +#include "../daemon/common.h" + +#define NETDATA_REGISTRY_COOKIE_NAME "netdata_registry_id" + +// initialize the registry +// should only happen when netdata starts +extern int registry_init(void); + +// free all data held by the registry +// should only happen when netdata exits +extern void registry_free(void); + +// HTTP requests handled by the registry +extern int registry_request_access_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *name, time_t when); +extern int registry_request_delete_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *delete_url, time_t when); +extern int registry_request_search_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *request_machine, time_t when); +extern int registry_request_switch_json(RRDHOST *host, struct web_client *w, char *person_guid, char *machine_guid, char *url, char *new_person_guid, time_t when); +extern int registry_request_hello_json(RRDHOST *host, struct web_client *w); + +// update the registry config +extern void registry_update_cloud_base_url(); + +// update the registry monitoring charts +extern void registry_statistics(void); + +extern char *registry_get_this_machine_guid(void); +extern char *registry_get_mgmt_api_key(void); +extern char *registry_get_this_machine_hostname(void); + +extern int regenerate_guid(const char *guid, char *result); + +#endif /* NETDATA_REGISTRY_H */ diff --git a/registry/registry_db.c b/registry/registry_db.c new file mode 100644 index 0000000..d8e2bbd --- /dev/null +++ b/registry/registry_db.c @@ -0,0 +1,346 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +int registry_db_should_be_saved(void) { + debug(D_REGISTRY, "log entries %llu, max %llu", registry.log_count, registry.save_registry_every_entries); + return registry.log_count > registry.save_registry_every_entries; +} + +// ---------------------------------------------------------------------------- +// INTERNAL FUNCTIONS FOR SAVING REGISTRY OBJECTS + +static int registry_machine_save_url(void *entry, void *file) { + REGISTRY_MACHINE_URL *mu = entry; + FILE *fp = file; + + debug(D_REGISTRY, "Registry: registry_machine_save_url('%s')", mu->url->url); + + int ret = fprintf(fp, "V\t%08x\t%08x\t%08x\t%02x\t%s\n", + mu->first_t, + mu->last_t, + mu->usages, + mu->flags, + mu->url->url + ); + + // error handling is done at registry_db_save() + + return ret; +} + +static int registry_machine_save(void *entry, void *file) { + REGISTRY_MACHINE *m = entry; + FILE *fp = file; + + debug(D_REGISTRY, "Registry: registry_machine_save('%s')", m->guid); + + int ret = fprintf(fp, "M\t%08x\t%08x\t%08x\t%s\n", + m->first_t, + m->last_t, + m->usages, + m->guid + ); + + if(ret >= 0) { + int ret2 = dictionary_get_all(m->machine_urls, registry_machine_save_url, fp); + if(ret2 < 0) return ret2; + ret += ret2; + } + + // error handling is done at registry_db_save() + + return ret; +} + +static inline int registry_person_save_url(void *entry, void *file) { + REGISTRY_PERSON_URL *pu = entry; + FILE *fp = file; + + debug(D_REGISTRY, "Registry: registry_person_save_url('%s')", pu->url->url); + + int ret = fprintf(fp, "U\t%08x\t%08x\t%08x\t%02x\t%s\t%s\t%s\n", + pu->first_t, + pu->last_t, + pu->usages, + pu->flags, + pu->machine->guid, + pu->machine_name, + pu->url->url + ); + + // error handling is done at registry_db_save() + + return ret; +} + +static inline int registry_person_save(void *entry, void *file) { + REGISTRY_PERSON *p = entry; + FILE *fp = file; + + debug(D_REGISTRY, "Registry: registry_person_save('%s')", p->guid); + + int ret = fprintf(fp, "P\t%08x\t%08x\t%08x\t%s\n", + p->first_t, + p->last_t, + p->usages, + p->guid + ); + + if(ret >= 0) { + //int ret2 = dictionary_get_all(p->person_urls, registry_person_save_url, fp); + int ret2 = avl_traverse(&p->person_urls, registry_person_save_url, fp); + if (ret2 < 0) return ret2; + ret += ret2; + } + + // error handling is done at registry_db_save() + + return ret; +} + +// ---------------------------------------------------------------------------- +// SAVE THE REGISTRY DATABASE + +int registry_db_save(void) { + if(unlikely(!registry.enabled)) + return -1; + + if(unlikely(!registry_db_should_be_saved())) + return -2; + + error_log_limit_unlimited(); + + char tmp_filename[FILENAME_MAX + 1]; + char old_filename[FILENAME_MAX + 1]; + + snprintfz(old_filename, FILENAME_MAX, "%s.old", registry.db_filename); + snprintfz(tmp_filename, FILENAME_MAX, "%s.tmp", registry.db_filename); + + debug(D_REGISTRY, "Registry: Creating file '%s'", tmp_filename); + FILE *fp = fopen(tmp_filename, "w"); + if(!fp) { + error("Registry: Cannot create file: %s", tmp_filename); + error_log_limit_reset(); + return -1; + } + + // dictionary_get_all() has its own locking, so this is safe to do + + debug(D_REGISTRY, "Saving all machines"); + int bytes1 = dictionary_get_all(registry.machines, registry_machine_save, fp); + if(bytes1 < 0) { + error("Registry: Cannot save registry machines - return value %d", bytes1); + fclose(fp); + error_log_limit_reset(); + return bytes1; + } + debug(D_REGISTRY, "Registry: saving machines took %d bytes", bytes1); + + debug(D_REGISTRY, "Saving all persons"); + int bytes2 = dictionary_get_all(registry.persons, registry_person_save, fp); + if(bytes2 < 0) { + error("Registry: Cannot save registry persons - return value %d", bytes2); + fclose(fp); + error_log_limit_reset(); + return bytes2; + } + debug(D_REGISTRY, "Registry: saving persons took %d bytes", bytes2); + + // save the totals + fprintf(fp, "T\t%016llx\t%016llx\t%016llx\t%016llx\t%016llx\t%016llx\n", + registry.persons_count, + registry.machines_count, + registry.usages_count + 1, // this is required - it is lost on db rotation + registry.urls_count, + registry.persons_urls_count, + registry.machines_urls_count + ); + + fclose(fp); + + errno = 0; + + // remove the .old db + debug(D_REGISTRY, "Registry: Removing old db '%s'", old_filename); + if(unlink(old_filename) == -1 && errno != ENOENT) + error("Registry: cannot remove old registry file '%s'", old_filename); + + // rename the db to .old + debug(D_REGISTRY, "Registry: Link current db '%s' to .old: '%s'", registry.db_filename, old_filename); + if(link(registry.db_filename, old_filename) == -1 && errno != ENOENT) + error("Registry: cannot move file '%s' to '%s'. Saving registry DB failed!", registry.db_filename, old_filename); + + else { + // remove the database (it is saved in .old) + debug(D_REGISTRY, "Registry: removing db '%s'", registry.db_filename); + if (unlink(registry.db_filename) == -1 && errno != ENOENT) + error("Registry: cannot remove old registry file '%s'", registry.db_filename); + + // move the .tmp to make it active + debug(D_REGISTRY, "Registry: linking tmp db '%s' to active db '%s'", tmp_filename, registry.db_filename); + if (link(tmp_filename, registry.db_filename) == -1) { + error("Registry: cannot move file '%s' to '%s'. Saving registry DB failed!", tmp_filename, + registry.db_filename); + + // move the .old back + debug(D_REGISTRY, "Registry: linking old db '%s' to active db '%s'", old_filename, registry.db_filename); + if(link(old_filename, registry.db_filename) == -1) + error("Registry: cannot move file '%s' to '%s'. Recovering the old registry DB failed!", old_filename, registry.db_filename); + } + else { + debug(D_REGISTRY, "Registry: removing tmp db '%s'", tmp_filename); + if(unlink(tmp_filename) == -1) + error("Registry: cannot remove tmp registry file '%s'", tmp_filename); + + // it has been moved successfully + // discard the current registry log + registry_log_recreate(); + registry.log_count = 0; + } + } + + // continue operations + error_log_limit_reset(); + + return -1; +} + +// ---------------------------------------------------------------------------- +// LOAD THE REGISTRY DATABASE + +size_t registry_db_load(void) { + char *s, buf[4096 + 1]; + REGISTRY_PERSON *p = NULL; + REGISTRY_MACHINE *m = NULL; + REGISTRY_URL *u = NULL; + size_t line = 0; + + debug(D_REGISTRY, "Registry: loading active db from: '%s'", registry.db_filename); + FILE *fp = fopen(registry.db_filename, "r"); + if(!fp) { + error("Registry: cannot open registry file: '%s'", registry.db_filename); + return 0; + } + + size_t len = 0; + buf[4096] = '\0'; + while((s = fgets_trim_len(buf, 4096, fp, &len))) { + line++; + + debug(D_REGISTRY, "Registry: read line %zu to length %zu: %s", line, len, s); + switch(*s) { + case 'T': // totals + if(unlikely(len != 103 || s[1] != '\t' || s[18] != '\t' || s[35] != '\t' || s[52] != '\t' || s[69] != '\t' || s[86] != '\t' || s[103] != '\0')) { + error("Registry totals line %zu is wrong (len = %zu).", line, len); + continue; + } + registry.persons_count = strtoull(&s[2], NULL, 16); + registry.machines_count = strtoull(&s[19], NULL, 16); + registry.usages_count = strtoull(&s[36], NULL, 16); + registry.urls_count = strtoull(&s[53], NULL, 16); + registry.persons_urls_count = strtoull(&s[70], NULL, 16); + registry.machines_urls_count = strtoull(&s[87], NULL, 16); + break; + + case 'P': // person + m = NULL; + // verify it is valid + if(unlikely(len != 65 || s[1] != '\t' || s[10] != '\t' || s[19] != '\t' || s[28] != '\t' || s[65] != '\0')) { + error("Registry person line %zu is wrong (len = %zu).", line, len); + continue; + } + + s[1] = s[10] = s[19] = s[28] = '\0'; + p = registry_person_allocate(&s[29], strtoul(&s[2], NULL, 16)); + p->last_t = (uint32_t)strtoul(&s[11], NULL, 16); + p->usages = (uint32_t)strtoul(&s[20], NULL, 16); + debug(D_REGISTRY, "Registry loaded person '%s', first: %u, last: %u, usages: %u", p->guid, p->first_t, p->last_t, p->usages); + break; + + case 'M': // machine + p = NULL; + // verify it is valid + if(unlikely(len != 65 || s[1] != '\t' || s[10] != '\t' || s[19] != '\t' || s[28] != '\t' || s[65] != '\0')) { + error("Registry person line %zu is wrong (len = %zu).", line, len); + continue; + } + + s[1] = s[10] = s[19] = s[28] = '\0'; + m = registry_machine_allocate(&s[29], strtoul(&s[2], NULL, 16)); + m->last_t = (uint32_t)strtoul(&s[11], NULL, 16); + m->usages = (uint32_t)strtoul(&s[20], NULL, 16); + debug(D_REGISTRY, "Registry loaded machine '%s', first: %u, last: %u, usages: %u", m->guid, m->first_t, m->last_t, m->usages); + break; + + case 'U': // person URL + if(unlikely(!p)) { + error("Registry: ignoring line %zu, no person loaded: %s", line, s); + continue; + } + + // verify it is valid + if(len < 69 || s[1] != '\t' || s[10] != '\t' || s[19] != '\t' || s[28] != '\t' || s[31] != '\t' || s[68] != '\t') { + error("Registry person URL line %zu is wrong (len = %zu).", line, len); + continue; + } + + s[1] = s[10] = s[19] = s[28] = s[31] = s[68] = '\0'; + + // skip the name to find the url + char *url = &s[69]; + while(*url && *url != '\t') url++; + if(!*url) { + error("Registry person URL line %zu does not have a url.", line); + continue; + } + *url++ = '\0'; + + // u = registry_url_allocate_nolock(url, strlen(url)); + u = registry_url_get(url, strlen(url)); + + time_t first_t = strtoul(&s[2], NULL, 16); + + m = registry_machine_find(&s[32]); + if(!m) m = registry_machine_allocate(&s[32], first_t); + + REGISTRY_PERSON_URL *pu = registry_person_url_allocate(p, m, u, &s[69], strlen(&s[69]), first_t); + pu->last_t = (uint32_t)strtoul(&s[11], NULL, 16); + pu->usages = (uint32_t)strtoul(&s[20], NULL, 16); + pu->flags = (uint8_t)strtoul(&s[29], NULL, 16); + debug(D_REGISTRY, "Registry loaded person URL '%s' with name '%s' of machine '%s', first: %u, last: %u, usages: %u, flags: %02x", u->url, pu->machine_name, m->guid, pu->first_t, pu->last_t, pu->usages, pu->flags); + break; + + case 'V': // machine URL + if(unlikely(!m)) { + error("Registry: ignoring line %zu, no machine loaded: %s", line, s); + continue; + } + + // verify it is valid + if(len < 32 || s[1] != '\t' || s[10] != '\t' || s[19] != '\t' || s[28] != '\t' || s[31] != '\t') { + error("Registry person URL line %zu is wrong (len = %zu).", line, len); + continue; + } + + s[1] = s[10] = s[19] = s[28] = s[31] = '\0'; + // u = registry_url_allocate_nolock(&s[32], strlen(&s[32])); + u = registry_url_get(&s[32], strlen(&s[32])); + + REGISTRY_MACHINE_URL *mu = registry_machine_url_allocate(m, u, strtoul(&s[2], NULL, 16)); + mu->last_t = (uint32_t)strtoul(&s[11], NULL, 16); + mu->usages = (uint32_t)strtoul(&s[20], NULL, 16); + mu->flags = (uint8_t)strtoul(&s[29], NULL, 16); + debug(D_REGISTRY, "Registry loaded machine URL '%s', machine '%s', first: %u, last: %u, usages: %u, flags: %02x", u->url, m->guid, mu->first_t, mu->last_t, mu->usages, mu->flags); + break; + + default: + error("Registry: ignoring line %zu of filename '%s': %s.", line, registry.db_filename, s); + break; + } + } + fclose(fp); + + return line; +} diff --git a/registry/registry_init.c b/registry/registry_init.c new file mode 100644 index 0000000..ffdb83f --- /dev/null +++ b/registry/registry_init.c @@ -0,0 +1,147 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +int registry_init(void) { + char filename[FILENAME_MAX + 1]; + + // registry enabled? + if(web_server_mode != WEB_SERVER_MODE_NONE) { + registry.enabled = config_get_boolean(CONFIG_SECTION_REGISTRY, "enabled", 0); + } + else { + info("Registry is disabled - use the central netdata"); + config_set_boolean(CONFIG_SECTION_REGISTRY, "enabled", 0); + registry.enabled = 0; + } + + // pathnames + snprintfz(filename, FILENAME_MAX, "%s/registry", netdata_configured_varlib_dir); + registry.pathname = config_get(CONFIG_SECTION_REGISTRY, "registry db directory", filename); + if(mkdir(registry.pathname, 0770) == -1 && errno != EEXIST) + fatal("Cannot create directory '%s'.", registry.pathname); + + // filenames + snprintfz(filename, FILENAME_MAX, "%s/netdata.public.unique.id", registry.pathname); + registry.machine_guid_filename = config_get(CONFIG_SECTION_REGISTRY, "netdata unique id file", filename); + + snprintfz(filename, FILENAME_MAX, "%s/registry.db", registry.pathname); + registry.db_filename = config_get(CONFIG_SECTION_REGISTRY, "registry db file", filename); + + snprintfz(filename, FILENAME_MAX, "%s/registry-log.db", registry.pathname); + registry.log_filename = config_get(CONFIG_SECTION_REGISTRY, "registry log file", filename); + + // configuration options + registry.save_registry_every_entries = (unsigned long long)config_get_number(CONFIG_SECTION_REGISTRY, "registry save db every new entries", 1000000); + registry.persons_expiration = config_get_number(CONFIG_SECTION_REGISTRY, "registry expire idle persons days", 365) * 86400; + registry.registry_domain = config_get(CONFIG_SECTION_REGISTRY, "registry domain", ""); + registry.registry_to_announce = config_get(CONFIG_SECTION_REGISTRY, "registry to announce", "https://registry.my-netdata.io"); + registry.hostname = config_get(CONFIG_SECTION_REGISTRY, "registry hostname", netdata_configured_hostname); + registry.verify_cookies_redirects = config_get_boolean(CONFIG_SECTION_REGISTRY, "verify browser cookies support", 1); + + registry_update_cloud_base_url(); + setenv("NETDATA_REGISTRY_HOSTNAME", registry.hostname, 1); + setenv("NETDATA_REGISTRY_URL", registry.registry_to_announce, 1); + + registry.max_url_length = (size_t)config_get_number(CONFIG_SECTION_REGISTRY, "max URL length", 1024); + if(registry.max_url_length < 10) { + registry.max_url_length = 10; + config_set_number(CONFIG_SECTION_REGISTRY, "max URL length", (long long)registry.max_url_length); + } + + registry.max_name_length = (size_t)config_get_number(CONFIG_SECTION_REGISTRY, "max URL name length", 50); + if(registry.max_name_length < 10) { + registry.max_name_length = 10; + config_set_number(CONFIG_SECTION_REGISTRY, "max URL name length", (long long)registry.max_name_length); + } + + // initialize entries counters + registry.persons_count = 0; + registry.machines_count = 0; + registry.usages_count = 0; + registry.urls_count = 0; + registry.persons_urls_count = 0; + registry.machines_urls_count = 0; + + // initialize memory counters + registry.persons_memory = 0; + registry.machines_memory = 0; + registry.urls_memory = 0; + registry.persons_urls_memory = 0; + registry.machines_urls_memory = 0; + + // initialize locks + netdata_mutex_init(®istry.lock); + + // create dictionaries + registry.persons = dictionary_create(DICTIONARY_FLAGS); + registry.machines = dictionary_create(DICTIONARY_FLAGS); + avl_init(®istry.registry_urls_root_index, registry_url_compare); + + // load the registry database + if(registry.enabled) { + registry_log_open(); + registry_db_load(); + registry_log_load(); + + if(unlikely(registry_db_should_be_saved())) + registry_db_save(); + } + + return 0; +} + +void registry_free(void) { + if(!registry.enabled) return; + + // we need to destroy the dictionaries ourselves + // since the dictionaries use memory we allocated + + while(registry.persons->values_index.root) { + REGISTRY_PERSON *p = ((NAME_VALUE *)registry.persons->values_index.root)->value; + registry_person_del(p); + } + + while(registry.machines->values_index.root) { + REGISTRY_MACHINE *m = ((NAME_VALUE *)registry.machines->values_index.root)->value; + + // fprintf(stderr, "\nMACHINE: '%s', first: %u, last: %u, usages: %u\n", m->guid, m->first_t, m->last_t, m->usages); + + while(m->machine_urls->values_index.root) { + REGISTRY_MACHINE_URL *mu = ((NAME_VALUE *)m->machine_urls->values_index.root)->value; + + // fprintf(stderr, "\tURL: '%s', first: %u, last: %u, usages: %u, flags: 0x%02x\n", mu->url->url, mu->first_t, mu->last_t, mu->usages, mu->flags); + + //debug(D_REGISTRY, "Registry: destroying persons dictionary from url '%s'", mu->url->url); + //dictionary_destroy(mu->persons); + + debug(D_REGISTRY, "Registry: deleting url '%s' from person '%s'", mu->url->url, m->guid); + dictionary_del(m->machine_urls, mu->url->url); + + debug(D_REGISTRY, "Registry: unlinking url '%s' from machine", mu->url->url); + registry_url_unlink(mu->url); + + debug(D_REGISTRY, "Registry: freeing machine url"); + freez(mu); + } + + debug(D_REGISTRY, "Registry: deleting machine '%s' from machines registry", m->guid); + dictionary_del(registry.machines, m->guid); + + debug(D_REGISTRY, "Registry: destroying URL dictionary of machine '%s'", m->guid); + dictionary_destroy(m->machine_urls); + + debug(D_REGISTRY, "Registry: freeing machine '%s'", m->guid); + freez(m); + } + + // and free the memory of remaining dictionary structures + + debug(D_REGISTRY, "Registry: destroying persons dictionary"); + dictionary_destroy(registry.persons); + + debug(D_REGISTRY, "Registry: destroying machines dictionary"); + dictionary_destroy(registry.machines); +} + diff --git a/registry/registry_internals.c b/registry/registry_internals.c new file mode 100644 index 0000000..3de6dd1 --- /dev/null +++ b/registry/registry_internals.c @@ -0,0 +1,332 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +struct registry registry; + +// ---------------------------------------------------------------------------- +// common functions + +// parse a GUID and re-generated to be always lower case +// this is used as a protection against the variations of GUIDs +int regenerate_guid(const char *guid, char *result) { + uuid_t uuid; + if(unlikely(uuid_parse(guid, uuid) == -1)) { + info("Registry: GUID '%s' is not a valid GUID.", guid); + return -1; + } + else { + uuid_unparse_lower(uuid, result); + +#ifdef NETDATA_INTERNAL_CHECKS + if(strcmp(guid, result) != 0) + info("GUID '%s' and re-generated GUID '%s' differ!", guid, result); +#endif /* NETDATA_INTERNAL_CHECKS */ + } + + return 0; +} + +// make sure the names of the machines / URLs do not contain any tabs +// (which are used as our separator in the database files) +// and are properly trimmed (before and after) +static inline char *registry_fix_machine_name(char *name, size_t *len) { + char *s = name?name:""; + + // skip leading spaces + while(*s && isspace(*s)) s++; + + // make sure all spaces are a SPACE + char *t = s; + while(*t) { + if(unlikely(isspace(*t))) + *t = ' '; + + t++; + } + + // remove trailing spaces + while(--t >= s) { + if(*t == ' ') + *t = '\0'; + else + break; + } + t++; + + if(likely(len)) + *len = (t - s); + + return s; +} + +static inline char *registry_fix_url(char *url, size_t *len) { + size_t l = 0; + char *s = registry_fix_machine_name(url, &l); + + // protection from too big URLs + if(l > registry.max_url_length) { + l = registry.max_url_length; + s[l] = '\0'; + } + + if(len) *len = l; + return s; +} + + +// ---------------------------------------------------------------------------- +// HELPERS + +// verify the person, the machine and the URL exist in our DB +REGISTRY_PERSON_URL *registry_verify_request(char *person_guid, char *machine_guid, char *url, REGISTRY_PERSON **pp, REGISTRY_MACHINE **mm) { + char pbuf[GUID_LEN + 1], mbuf[GUID_LEN + 1]; + + if(!person_guid || !*person_guid || !machine_guid || !*machine_guid || !url || !*url) { + info("Registry Request Verification: invalid request! person: '%s', machine '%s', url '%s'", person_guid?person_guid:"UNSET", machine_guid?machine_guid:"UNSET", url?url:"UNSET"); + return NULL; + } + + // normalize the url + url = registry_fix_url(url, NULL); + + // make sure the person GUID is valid + if(regenerate_guid(person_guid, pbuf) == -1) { + info("Registry Request Verification: invalid person GUID, person: '%s', machine '%s', url '%s'", person_guid, machine_guid, url); + return NULL; + } + person_guid = pbuf; + + // make sure the machine GUID is valid + if(regenerate_guid(machine_guid, mbuf) == -1) { + info("Registry Request Verification: invalid machine GUID, person: '%s', machine '%s', url '%s'", person_guid, machine_guid, url); + return NULL; + } + machine_guid = mbuf; + + // make sure the machine exists + REGISTRY_MACHINE *m = registry_machine_find(machine_guid); + if(!m) { + info("Registry Request Verification: machine not found, person: '%s', machine '%s', url '%s'", person_guid, machine_guid, url); + return NULL; + } + if(mm) *mm = m; + + // make sure the person exist + REGISTRY_PERSON *p = registry_person_find(person_guid); + if(!p) { + info("Registry Request Verification: person not found, person: '%s', machine '%s', url '%s'", person_guid, machine_guid, url); + return NULL; + } + if(pp) *pp = p; + + REGISTRY_PERSON_URL *pu = registry_person_url_index_find(p, url); + if(!pu) { + info("Registry Request Verification: URL not found for person, person: '%s', machine '%s', url '%s'", person_guid, machine_guid, url); + return NULL; + } + //else if (pu->machine != m) { + // info("Registry Request Verification: Machine mismatch: person: '%s', machine requested='%s' <> loaded='%s', url '%s'", person_guid, machine_guid, pu->machine->guid, url); + // return NULL; + //} + + return pu; +} + + +// ---------------------------------------------------------------------------- +// REGISTRY REQUESTS + +REGISTRY_PERSON *registry_request_access(char *person_guid, char *machine_guid, char *url, char *name, time_t when) { + debug(D_REGISTRY, "registry_request_access('%s', '%s', '%s'): NEW REQUEST", (person_guid)?person_guid:"", machine_guid, url); + + REGISTRY_MACHINE *m = registry_machine_get(machine_guid, when); + if(!m) return NULL; + + // make sure the name is valid + size_t namelen; + name = registry_fix_machine_name(name, &namelen); + + size_t urllen; + url = registry_fix_url(url, &urllen); + + REGISTRY_PERSON *p = registry_person_get(person_guid, when); + + REGISTRY_URL *u = registry_url_get(url, urllen); + registry_person_link_to_url(p, m, u, name, namelen, when); + registry_machine_link_to_url(m, u, when); + + registry_log('A', p, m, u, name); + + registry.usages_count++; + + return p; +} + +REGISTRY_PERSON *registry_request_delete(char *person_guid, char *machine_guid, char *url, char *delete_url, time_t when) { + (void) when; + + REGISTRY_PERSON *p = NULL; + REGISTRY_MACHINE *m = NULL; + REGISTRY_PERSON_URL *pu = registry_verify_request(person_guid, machine_guid, url, &p, &m); + if(!pu || !p || !m) return NULL; + + // normalize the url + delete_url = registry_fix_url(delete_url, NULL); + + // make sure the user is not deleting the url it uses + /* + if(!strcmp(delete_url, pu->url->url)) { + info("Registry Delete Request: delete URL is the one currently accessed, person: '%s', machine '%s', url '%s', delete url '%s'" + , p->guid, m->guid, pu->url->url, delete_url); + return NULL; + } + */ + + REGISTRY_PERSON_URL *dpu = registry_person_url_index_find(p, delete_url); + if(!dpu) { + info("Registry Delete Request: URL not found for person: '%s', machine '%s', url '%s', delete url '%s'", p->guid + , m->guid, pu->url->url, delete_url); + return NULL; + } + + registry_log('D', p, m, pu->url, dpu->url->url); + registry_person_unlink_from_url(p, dpu); + + return p; +} + + +// a structure to pass to the dictionary_get_all() callback handler +struct machine_request_callback_data { + REGISTRY_MACHINE *find_this_machine; + REGISTRY_PERSON_URL *result; +}; + +// the callback function +// this will be run for every PERSON_URL of this PERSON +static int machine_request_callback(void *entry, void *data) { + REGISTRY_PERSON_URL *mypu = (REGISTRY_PERSON_URL *)entry; + struct machine_request_callback_data *myrdata = (struct machine_request_callback_data *)data; + + if(mypu->machine == myrdata->find_this_machine) { + myrdata->result = mypu; + return -1; // this will also stop the walk through + } + + return 0; // continue +} + +REGISTRY_MACHINE *registry_request_machine(char *person_guid, char *machine_guid, char *url, char *request_machine, time_t when) { + (void)when; + + char mbuf[GUID_LEN + 1]; + + REGISTRY_PERSON *p = NULL; + REGISTRY_MACHINE *m = NULL; + REGISTRY_PERSON_URL *pu = registry_verify_request(person_guid, machine_guid, url, &p, &m); + if(!pu || !p || !m) return NULL; + + // make sure the machine GUID is valid + if(regenerate_guid(request_machine, mbuf) == -1) { + info("Registry Machine URLs request: invalid machine GUID, person: '%s', machine '%s', url '%s', request machine '%s'", p->guid, m->guid, pu->url->url, request_machine); + return NULL; + } + request_machine = mbuf; + + // make sure the machine exists + m = registry_machine_find(request_machine); + if(!m) { + info("Registry Machine URLs request: machine not found, person: '%s', machine '%s', url '%s', request machine '%s'", p->guid, machine_guid, pu->url->url, request_machine); + return NULL; + } + + // Verify the user has in the past accessed this machine + // We will walk through the PERSON_URLs to find the machine + // linking to our machine + + // a structure to pass to the dictionary_get_all() callback handler + struct machine_request_callback_data rdata = { m, NULL }; + + // request a walk through on the dictionary + avl_traverse(&p->person_urls, machine_request_callback, &rdata); + + if(rdata.result) + return m; + + return NULL; +} + + +// ---------------------------------------------------------------------------- +// REGISTRY THIS MACHINE UNIQUE ID + +static inline int is_machine_guid_blacklisted(const char *guid) { + // these are machine GUIDs that have been included in distribution packages. + // we blacklist them here, so that the next version of netdata will generate + // new ones. + + if(!strcmp(guid, "8a795b0c-2311-11e6-8563-000c295076a6") + || !strcmp(guid, "4aed1458-1c3e-11e6-a53f-000c290fc8f5") + ) { + error("Blacklisted machine GUID '%s' found.", guid); + return 1; + } + + return 0; +} + +char *registry_get_this_machine_hostname(void) { + return registry.hostname; +} + +char *registry_get_this_machine_guid(void) { + static char guid[GUID_LEN + 1] = ""; + + if(likely(guid[0])) + return guid; + + // read it from disk + int fd = open(registry.machine_guid_filename, O_RDONLY); + if(fd != -1) { + char buf[GUID_LEN + 1]; + if(read(fd, buf, GUID_LEN) != GUID_LEN) + error("Failed to read machine GUID from '%s'", registry.machine_guid_filename); + else { + buf[GUID_LEN] = '\0'; + if(regenerate_guid(buf, guid) == -1) { + error("Failed to validate machine GUID '%s' from '%s'. Ignoring it - this might mean this netdata will appear as duplicate in the registry.", + buf, registry.machine_guid_filename); + + guid[0] = '\0'; + } + else if(is_machine_guid_blacklisted(guid)) + guid[0] = '\0'; + } + close(fd); + } + + // generate a new one? + if(!guid[0]) { + uuid_t uuid; + + uuid_generate_time(uuid); + uuid_unparse_lower(uuid, guid); + guid[GUID_LEN] = '\0'; + + // save it + fd = open(registry.machine_guid_filename, O_WRONLY|O_CREAT|O_TRUNC, 444); + if(fd == -1) + fatal("Cannot create unique machine id file '%s'. Please fix this.", registry.machine_guid_filename); + + if(write(fd, guid, GUID_LEN) != GUID_LEN) + fatal("Cannot write the unique machine id file '%s'. Please fix this.", registry.machine_guid_filename); + + close(fd); + } + + setenv("NETDATA_REGISTRY_UNIQUE_ID", guid, 1); + + return guid; +} diff --git a/registry/registry_internals.h b/registry/registry_internals.h new file mode 100644 index 0000000..0eb83a4 --- /dev/null +++ b/registry/registry_internals.h @@ -0,0 +1,89 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#ifndef NETDATA_REGISTRY_INTERNALS_H_H +#define NETDATA_REGISTRY_INTERNALS_H_H 1 + +#include "registry.h" + +#define REGISTRY_URL_FLAGS_DEFAULT 0x00 +#define REGISTRY_URL_FLAGS_EXPIRED 0x01 + +#define DICTIONARY_FLAGS (DICTIONARY_FLAG_VALUE_LINK_DONT_CLONE | DICTIONARY_FLAG_NAME_LINK_DONT_CLONE | DICTIONARY_FLAG_SINGLE_THREADED) + +// ---------------------------------------------------------------------------- +// COMMON structures + +struct registry { + int enabled; + + // entries counters / statistics + unsigned long long persons_count; + unsigned long long machines_count; + unsigned long long usages_count; + unsigned long long urls_count; + unsigned long long persons_urls_count; + unsigned long long machines_urls_count; + unsigned long long log_count; + + // memory counters / statistics + unsigned long long persons_memory; + unsigned long long machines_memory; + unsigned long long urls_memory; + unsigned long long persons_urls_memory; + unsigned long long machines_urls_memory; + + // configuration + unsigned long long save_registry_every_entries; + char *registry_domain; + char *hostname; + char *registry_to_announce; + char *cloud_base_url; + time_t persons_expiration; // seconds to expire idle persons + int verify_cookies_redirects; + + size_t max_url_length; + size_t max_name_length; + + // file/path names + char *pathname; + char *db_filename; + char *log_filename; + char *machine_guid_filename; + + // open files + FILE *log_fp; + + // the database + DICTIONARY *persons; // dictionary of REGISTRY_PERSON *, with key the REGISTRY_PERSON.guid + DICTIONARY *machines; // dictionary of REGISTRY_MACHINE *, with key the REGISTRY_MACHINE.guid + + avl_tree_type registry_urls_root_index; + + netdata_mutex_t lock; +}; + +#include "registry_url.h" +#include "registry_machine.h" +#include "registry_person.h" +#include "registry.h" + +extern struct registry registry; + +// REGISTRY LOW-LEVEL REQUESTS (in registry-internals.c) +extern REGISTRY_PERSON *registry_request_access(char *person_guid, char *machine_guid, char *url, char *name, time_t when); +extern REGISTRY_PERSON *registry_request_delete(char *person_guid, char *machine_guid, char *url, char *delete_url, time_t when); +extern REGISTRY_MACHINE *registry_request_machine(char *person_guid, char *machine_guid, char *url, char *request_machine, time_t when); + +// REGISTRY LOG (in registry_log.c) +extern void registry_log(char action, REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name); +extern int registry_log_open(void); +extern void registry_log_close(void); +extern void registry_log_recreate(void); +extern ssize_t registry_log_load(void); + +// REGISTRY DB (in registry_db.c) +extern int registry_db_save(void); +extern size_t registry_db_load(void); +extern int registry_db_should_be_saved(void); + +#endif //NETDATA_REGISTRY_INTERNALS_H_H diff --git a/registry/registry_log.c b/registry/registry_log.c new file mode 100644 index 0000000..e0e58ed --- /dev/null +++ b/registry/registry_log.c @@ -0,0 +1,136 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +void registry_log(char action, REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name) { + if(likely(registry.log_fp)) { + if(unlikely(fprintf(registry.log_fp, "%c\t%08x\t%s\t%s\t%s\t%s\n", + action, + p->last_t, + p->guid, + m->guid, + name, + u->url) < 0)) + error("Registry: failed to save log. Registry data may be lost in case of abnormal restart."); + + // we increase the counter even on failures + // so that the registry will be saved periodically + registry.log_count++; + + // this must be outside the log_lock(), or a deadlock will happen. + // registry_db_save() checks the same inside the log_lock, so only + // one thread will save the db + if(unlikely(registry_db_should_be_saved())) + registry_db_save(); + } +} + +int registry_log_open(void) { + if(registry.log_fp) + fclose(registry.log_fp); + + registry.log_fp = fopen(registry.log_filename, "a"); + if(registry.log_fp) { + if (setvbuf(registry.log_fp, NULL, _IOLBF, 0) != 0) + error("Cannot set line buffering on registry log file."); + return 0; + } + + error("Cannot open registry log file '%s'. Registry data will be lost in case of netdata or server crash.", registry.log_filename); + return -1; +} + +void registry_log_close(void) { + if(registry.log_fp) { + fclose(registry.log_fp); + registry.log_fp = NULL; + } +} + +void registry_log_recreate(void) { + if(registry.log_fp != NULL) { + registry_log_close(); + + // open it with truncate + registry.log_fp = fopen(registry.log_filename, "w"); + if(registry.log_fp) fclose(registry.log_fp); + else error("Cannot truncate registry log '%s'", registry.log_filename); + + registry.log_fp = NULL; + registry_log_open(); + } +} + +ssize_t registry_log_load(void) { + ssize_t line = -1; + + // closing the log is required here + // otherwise we will append to it the values we read + registry_log_close(); + + debug(D_REGISTRY, "Registry: loading active db from: %s", registry.log_filename); + FILE *fp = fopen(registry.log_filename, "r"); + if(!fp) + error("Registry: cannot open registry file: %s", registry.log_filename); + else { + char *s, buf[4096 + 1]; + line = 0; + size_t len = 0; + + while ((s = fgets_trim_len(buf, 4096, fp, &len))) { + line++; + + switch (s[0]) { + case 'A': // accesses + case 'D': // deletes + + // verify it is valid + if (unlikely(len < 85 || s[1] != '\t' || s[10] != '\t' || s[47] != '\t' || s[84] != '\t')) { + error("Registry: log line %zd is wrong (len = %zu).", line, len); + continue; + } + s[1] = s[10] = s[47] = s[84] = '\0'; + + // get the variables + time_t when = strtoul(&s[2], NULL, 16); + char *person_guid = &s[11]; + char *machine_guid = &s[48]; + char *name = &s[85]; + + // skip the name to find the url + char *url = name; + while(*url && *url != '\t') url++; + if(!*url) { + error("Registry: log line %zd does not have a url.", line); + continue; + } + *url++ = '\0'; + + // make sure the person exists + // without this, a new person guid will be created + REGISTRY_PERSON *p = registry_person_find(person_guid); + if(!p) p = registry_person_allocate(person_guid, when); + + if(s[0] == 'A') + registry_request_access(p->guid, machine_guid, url, name, when); + else + registry_request_delete(p->guid, machine_guid, url, name, when); + + registry.log_count++; + break; + + default: + error("Registry: ignoring line %zd of filename '%s': %s.", line, registry.log_filename, s); + break; + } + } + + fclose(fp); + } + + // open the log again + registry_log_open(); + + return line; +} diff --git a/registry/registry_machine.c b/registry/registry_machine.c new file mode 100644 index 0000000..8dbeb8e --- /dev/null +++ b/registry/registry_machine.c @@ -0,0 +1,104 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +// ---------------------------------------------------------------------------- +// MACHINE + +REGISTRY_MACHINE *registry_machine_find(const char *machine_guid) { + debug(D_REGISTRY, "Registry: registry_machine_find('%s')", machine_guid); + return dictionary_get(registry.machines, machine_guid); +} + +REGISTRY_MACHINE_URL *registry_machine_url_allocate(REGISTRY_MACHINE *m, REGISTRY_URL *u, time_t when) { + debug(D_REGISTRY, "registry_machine_url_allocate('%s', '%s'): allocating %zu bytes", m->guid, u->url, sizeof(REGISTRY_MACHINE_URL)); + + REGISTRY_MACHINE_URL *mu = mallocz(sizeof(REGISTRY_MACHINE_URL)); + + mu->first_t = mu->last_t = (uint32_t)when; + mu->usages = 1; + mu->url = u; + mu->flags = REGISTRY_URL_FLAGS_DEFAULT; + + registry.machines_urls_memory += sizeof(REGISTRY_MACHINE_URL); + + debug(D_REGISTRY, "registry_machine_url_allocate('%s', '%s'): indexing URL in machine", m->guid, u->url); + dictionary_set(m->machine_urls, u->url, mu, sizeof(REGISTRY_MACHINE_URL)); + + registry_url_link(u); + + return mu; +} + +REGISTRY_MACHINE *registry_machine_allocate(const char *machine_guid, time_t when) { + debug(D_REGISTRY, "Registry: registry_machine_allocate('%s'): creating new machine, sizeof(MACHINE)=%zu", machine_guid, sizeof(REGISTRY_MACHINE)); + + REGISTRY_MACHINE *m = mallocz(sizeof(REGISTRY_MACHINE)); + + strncpyz(m->guid, machine_guid, GUID_LEN); + + debug(D_REGISTRY, "Registry: registry_machine_allocate('%s'): creating dictionary of urls", machine_guid); + m->machine_urls = dictionary_create(DICTIONARY_FLAGS); + + m->first_t = m->last_t = (uint32_t)when; + m->usages = 0; + + registry.machines_memory += sizeof(REGISTRY_MACHINE); + + registry.machines_count++; + dictionary_set(registry.machines, m->guid, m, sizeof(REGISTRY_MACHINE)); + + return m; +} + +// 1. validate machine GUID +// 2. if it is valid, find it or create it and return it +// 3. if it is not valid, return NULL +REGISTRY_MACHINE *registry_machine_get(const char *machine_guid, time_t when) { + REGISTRY_MACHINE *m = NULL; + + if(likely(machine_guid && *machine_guid)) { + // validate it is a GUID + char buf[GUID_LEN + 1]; + if(unlikely(regenerate_guid(machine_guid, buf) == -1)) + info("Registry: machine guid '%s' is not a valid guid. Ignoring it.", machine_guid); + else { + machine_guid = buf; + m = registry_machine_find(machine_guid); + if(!m) m = registry_machine_allocate(machine_guid, when); + } + } + + return m; +} + + +// ---------------------------------------------------------------------------- +// LINKING OF OBJECTS + +REGISTRY_MACHINE_URL *registry_machine_link_to_url(REGISTRY_MACHINE *m, REGISTRY_URL *u, time_t when) { + debug(D_REGISTRY, "registry_machine_link_to_url('%s', '%s'): searching for URL in machine", m->guid, u->url); + + REGISTRY_MACHINE_URL *mu = dictionary_get(m->machine_urls, u->url); + if(!mu) { + debug(D_REGISTRY, "registry_machine_link_to_url('%s', '%s'): not found", m->guid, u->url); + mu = registry_machine_url_allocate(m, u, when); + registry.machines_urls_count++; + } + else { + debug(D_REGISTRY, "registry_machine_link_to_url('%s', '%s'): found", m->guid, u->url); + mu->usages++; + if(likely(mu->last_t < (uint32_t)when)) mu->last_t = (uint32_t)when; + } + + m->usages++; + if(likely(m->last_t < (uint32_t)when)) m->last_t = (uint32_t)when; + + if(mu->flags & REGISTRY_URL_FLAGS_EXPIRED) { + debug(D_REGISTRY, "registry_machine_link_to_url('%s', '%s'): accessing an expired URL.", m->guid, u->url); + mu->flags &= ~REGISTRY_URL_FLAGS_EXPIRED; + } + + return mu; +} diff --git a/registry/registry_machine.h b/registry/registry_machine.h new file mode 100644 index 0000000..77ab5aa --- /dev/null +++ b/registry/registry_machine.h @@ -0,0 +1,43 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#ifndef NETDATA_REGISTRY_MACHINE_H +#define NETDATA_REGISTRY_MACHINE_H 1 + +#include "registry_internals.h" + +// ---------------------------------------------------------------------------- +// MACHINE structures + +// For each MACHINE-URL pair we keep this +struct registry_machine_url { + REGISTRY_URL *url; // de-duplicated URL + + uint8_t flags; + + uint32_t first_t; // the first time we saw this + uint32_t last_t; // the last time we saw this + uint32_t usages; // how many times this has been accessed +}; +typedef struct registry_machine_url REGISTRY_MACHINE_URL; + +// A machine +struct registry_machine { + char guid[GUID_LEN + 1]; // the GUID + + uint32_t links; // the number of REGISTRY_PERSON_URL linked to this machine + + DICTIONARY *machine_urls; // MACHINE_URL * + + uint32_t first_t; // the first time we saw this + uint32_t last_t; // the last time we saw this + uint32_t usages; // how many times this has been accessed +}; +typedef struct registry_machine REGISTRY_MACHINE; + +extern REGISTRY_MACHINE *registry_machine_find(const char *machine_guid); +extern REGISTRY_MACHINE_URL *registry_machine_url_allocate(REGISTRY_MACHINE *m, REGISTRY_URL *u, time_t when); +extern REGISTRY_MACHINE *registry_machine_allocate(const char *machine_guid, time_t when); +extern REGISTRY_MACHINE *registry_machine_get(const char *machine_guid, time_t when); +extern REGISTRY_MACHINE_URL *registry_machine_link_to_url(REGISTRY_MACHINE *m, REGISTRY_URL *u, time_t when); + +#endif //NETDATA_REGISTRY_MACHINE_H diff --git a/registry/registry_person.c b/registry/registry_person.c new file mode 100644 index 0000000..268b0bd --- /dev/null +++ b/registry/registry_person.c @@ -0,0 +1,267 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +// ---------------------------------------------------------------------------- +// PERSON_URL INDEX + +int person_url_compare(void *a, void *b) { + register uint32_t hash1 = ((REGISTRY_PERSON_URL *)a)->url->hash; + register uint32_t hash2 = ((REGISTRY_PERSON_URL *)b)->url->hash; + + if(hash1 < hash2) return -1; + else if(hash1 > hash2) return 1; + else return strcmp(((REGISTRY_PERSON_URL *)a)->url->url, ((REGISTRY_PERSON_URL *)b)->url->url); +} + +inline REGISTRY_PERSON_URL *registry_person_url_index_find(REGISTRY_PERSON *p, const char *url) { + debug(D_REGISTRY, "Registry: registry_person_url_index_find('%s', '%s')", p->guid, url); + + char buf[sizeof(REGISTRY_URL) + strlen(url)]; + + REGISTRY_URL *u = (REGISTRY_URL *)&buf; + strcpy(u->url, url); + u->hash = simple_hash(u->url); + + REGISTRY_PERSON_URL tpu = { .url = u }; + + REGISTRY_PERSON_URL *pu = (REGISTRY_PERSON_URL *)avl_search(&p->person_urls, (void *)&tpu); + return pu; +} + +inline REGISTRY_PERSON_URL *registry_person_url_index_add(REGISTRY_PERSON *p, REGISTRY_PERSON_URL *pu) { + debug(D_REGISTRY, "Registry: registry_person_url_index_add('%s', '%s')", p->guid, pu->url->url); + REGISTRY_PERSON_URL *tpu = (REGISTRY_PERSON_URL *)avl_insert(&(p->person_urls), (avl *)(pu)); + if(tpu != pu) + error("Registry: registry_person_url_index_add('%s', '%s') already exists as '%s'", p->guid, pu->url->url, tpu->url->url); + + return tpu; +} + +inline REGISTRY_PERSON_URL *registry_person_url_index_del(REGISTRY_PERSON *p, REGISTRY_PERSON_URL *pu) { + debug(D_REGISTRY, "Registry: registry_person_url_index_del('%s', '%s')", p->guid, pu->url->url); + REGISTRY_PERSON_URL *tpu = (REGISTRY_PERSON_URL *)avl_remove(&(p->person_urls), (avl *)(pu)); + if(!tpu) + error("Registry: registry_person_url_index_del('%s', '%s') deleted nothing", p->guid, pu->url->url); + else if(tpu != pu) + error("Registry: registry_person_url_index_del('%s', '%s') deleted wrong URL '%s'", p->guid, pu->url->url, tpu->url->url); + + return tpu; +} + +// ---------------------------------------------------------------------------- +// PERSON_URL + +REGISTRY_PERSON_URL *registry_person_url_allocate(REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name, size_t namelen, time_t when) { + debug(D_REGISTRY, "registry_person_url_allocate('%s', '%s', '%s'): allocating %zu bytes", p->guid, m->guid, u->url, sizeof(REGISTRY_PERSON_URL) + namelen); + + // protection from too big names + if(namelen > registry.max_name_length) + namelen = registry.max_name_length; + + REGISTRY_PERSON_URL *pu = mallocz(sizeof(REGISTRY_PERSON_URL) + namelen); + + // a simple strcpy() should do the job + // but I prefer to be safe, since the caller specified urllen + strncpyz(pu->machine_name, name, namelen); + + pu->machine = m; + pu->first_t = pu->last_t = (uint32_t)when; + pu->usages = 1; + pu->url = u; + pu->flags = REGISTRY_URL_FLAGS_DEFAULT; + m->links++; + + registry.persons_urls_memory += sizeof(REGISTRY_PERSON_URL) + namelen; + + debug(D_REGISTRY, "registry_person_url_allocate('%s', '%s', '%s'): indexing URL in person", p->guid, m->guid, u->url); + REGISTRY_PERSON_URL *tpu = registry_person_url_index_add(p, pu); + if(tpu != pu) { + error("Registry: Attempted to add duplicate person url '%s' with name '%s' to person '%s'", u->url, name, p->guid); + freez(pu); + pu = tpu; + } + else + registry_url_link(u); + + return pu; +} + +void registry_person_url_free(REGISTRY_PERSON *p, REGISTRY_PERSON_URL *pu) { + debug(D_REGISTRY, "registry_person_url_free('%s', '%s')", p->guid, pu->url->url); + + REGISTRY_PERSON_URL *tpu = registry_person_url_index_del(p, pu); + if(tpu) { + registry_url_unlink(tpu->url); + tpu->machine->links--; + registry.persons_urls_memory -= sizeof(REGISTRY_PERSON_URL) + strlen(tpu->machine_name); + freez(tpu); + } +} + +// this function is needed to change the name of a PERSON_URL +REGISTRY_PERSON_URL *registry_person_url_reallocate(REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name, size_t namelen, time_t when, REGISTRY_PERSON_URL *pu) { + debug(D_REGISTRY, "registry_person_url_reallocate('%s', '%s', '%s'): allocating %zu bytes", p->guid, m->guid, u->url, sizeof(REGISTRY_PERSON_URL) + namelen); + + // keep a backup + REGISTRY_PERSON_URL pu2 = { + .first_t = pu->first_t, + .last_t = pu->last_t, + .usages = pu->usages, + .flags = pu->flags, + .machine = pu->machine, + .machine_name = "" + }; + + // remove the existing one from the index + registry_person_url_free(p, pu); + pu = &pu2; + + // allocate a new one + REGISTRY_PERSON_URL *tpu = registry_person_url_allocate(p, m, u, name, namelen, when); + tpu->first_t = pu->first_t; + tpu->last_t = pu->last_t; + tpu->usages = pu->usages; + tpu->flags = pu->flags; + + return tpu; +} + + +// ---------------------------------------------------------------------------- +// PERSON + +REGISTRY_PERSON *registry_person_find(const char *person_guid) { + debug(D_REGISTRY, "Registry: registry_person_find('%s')", person_guid); + return dictionary_get(registry.persons, person_guid); +} + +REGISTRY_PERSON *registry_person_allocate(const char *person_guid, time_t when) { + debug(D_REGISTRY, "Registry: registry_person_allocate('%s'): allocating new person, sizeof(PERSON)=%zu", (person_guid)?person_guid:"", sizeof(REGISTRY_PERSON)); + + REGISTRY_PERSON *p = mallocz(sizeof(REGISTRY_PERSON)); + if(!person_guid) { + for(;;) { + uuid_t uuid; + uuid_generate(uuid); + uuid_unparse_lower(uuid, p->guid); + + debug(D_REGISTRY, "Registry: Checking if the generated person guid '%s' is unique", p->guid); + if (!dictionary_get(registry.persons, p->guid)) { + debug(D_REGISTRY, "Registry: generated person guid '%s' is unique", p->guid); + break; + } + else + info("Registry: generated person guid '%s' found in the registry. Retrying...", p->guid); + } + } + else + strncpyz(p->guid, person_guid, GUID_LEN); + + debug(D_REGISTRY, "Registry: registry_person_allocate('%s'): creating dictionary of urls", p->guid); + avl_init(&p->person_urls, person_url_compare); + + p->first_t = p->last_t = (uint32_t)when; + p->usages = 0; + + registry.persons_memory += sizeof(REGISTRY_PERSON); + + registry.persons_count++; + dictionary_set(registry.persons, p->guid, p, sizeof(REGISTRY_PERSON)); + + return p; +} + + +// 1. validate person GUID +// 2. if it is valid, find it +// 3. if it is not valid, create a new one +// 4. return it +REGISTRY_PERSON *registry_person_get(const char *person_guid, time_t when) { + debug(D_REGISTRY, "Registry: registry_person_get('%s'): creating dictionary of urls", person_guid); + + REGISTRY_PERSON *p = NULL; + + if(person_guid && *person_guid) { + char buf[GUID_LEN + 1]; + // validate it is a GUID + if(unlikely(regenerate_guid(person_guid, buf) == -1)) + info("Registry: person guid '%s' is not a valid guid. Ignoring it.", person_guid); + else { + person_guid = buf; + p = registry_person_find(person_guid); + } + } + + if(!p) p = registry_person_allocate(NULL, when); + + return p; +} + +void registry_person_del(REGISTRY_PERSON *p) { + debug(D_REGISTRY, "Registry: registry_person_del('%s'): creating dictionary of urls", p->guid); + + while(p->person_urls.root) + registry_person_unlink_from_url(p, (REGISTRY_PERSON_URL *)p->person_urls.root); + + debug(D_REGISTRY, "Registry: deleting person '%s' from persons registry", p->guid); + dictionary_del(registry.persons, p->guid); + + debug(D_REGISTRY, "Registry: freeing person '%s'", p->guid); + freez(p); +} + +// ---------------------------------------------------------------------------- +// LINKING OF OBJECTS + +REGISTRY_PERSON_URL *registry_person_link_to_url(REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name, size_t namelen, time_t when) { + debug(D_REGISTRY, "registry_person_link_to_url('%s', '%s', '%s'): searching for URL in person", p->guid, m->guid, u->url); + + REGISTRY_PERSON_URL *pu = registry_person_url_index_find(p, u->url); + if(!pu) { + debug(D_REGISTRY, "registry_person_link_to_url('%s', '%s', '%s'): not found", p->guid, m->guid, u->url); + pu = registry_person_url_allocate(p, m, u, name, namelen, when); + registry.persons_urls_count++; + } + else { + debug(D_REGISTRY, "registry_person_link_to_url('%s', '%s', '%s'): found", p->guid, m->guid, u->url); + pu->usages++; + if(likely(pu->last_t < (uint32_t)when)) pu->last_t = (uint32_t)when; + + if(pu->machine != m) { + REGISTRY_MACHINE_URL *mu = dictionary_get(pu->machine->machine_urls, u->url); + if(mu) { + debug(D_REGISTRY, "registry_person_link_to_url('%s', '%s', '%s'): URL switched machines (old was '%s') - expiring it from previous machine.", + p->guid, m->guid, u->url, pu->machine->guid); + mu->flags |= REGISTRY_URL_FLAGS_EXPIRED; + } + else { + debug(D_REGISTRY, "registry_person_link_to_url('%s', '%s', '%s'): URL switched machines (old was '%s') - but the URL is not linked to the old machine.", + p->guid, m->guid, u->url, pu->machine->guid); + } + + pu->machine->links--; + pu->machine = m; + } + + if(strcmp(pu->machine_name, name) != 0) { + // the name of the PERSON_URL has changed ! + pu = registry_person_url_reallocate(p, m, u, name, namelen, when, pu); + } + } + + p->usages++; + if(likely(p->last_t < (uint32_t)when)) p->last_t = (uint32_t)when; + + if(pu->flags & REGISTRY_URL_FLAGS_EXPIRED) { + debug(D_REGISTRY, "registry_person_link_to_url('%s', '%s', '%s'): accessing an expired URL. Re-enabling URL.", p->guid, m->guid, u->url); + pu->flags &= ~REGISTRY_URL_FLAGS_EXPIRED; + } + + return pu; +} + +void registry_person_unlink_from_url(REGISTRY_PERSON *p, REGISTRY_PERSON_URL *pu) { + registry_person_url_free(p, pu); +} diff --git a/registry/registry_person.h b/registry/registry_person.h new file mode 100644 index 0000000..9a4aa95 --- /dev/null +++ b/registry/registry_person.h @@ -0,0 +1,62 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#ifndef NETDATA_REGISTRY_PERSON_H +#define NETDATA_REGISTRY_PERSON_H 1 + +#include "registry_internals.h" + +// ---------------------------------------------------------------------------- +// PERSON structures + +// for each PERSON-URL pair we keep this +struct registry_person_url { + avl avl; // binary tree node + + REGISTRY_URL *url; // de-duplicated URL + REGISTRY_MACHINE *machine; // link the MACHINE of this URL + + uint8_t flags; + + uint32_t first_t; // the first time we saw this + uint32_t last_t; // the last time we saw this + uint32_t usages; // how many times this has been accessed + + char machine_name[1]; // the name of the machine, as known by the user + // dynamically allocated to fit properly +}; +typedef struct registry_person_url REGISTRY_PERSON_URL; + +// A person +struct registry_person { + char guid[GUID_LEN + 1]; // the person GUID + + avl_tree_type person_urls; // dictionary of PERSON_URLs + + uint32_t first_t; // the first time we saw this + uint32_t last_t; // the last time we saw this + uint32_t usages; // how many times this has been accessed + + //uint32_t flags; + //char *email; +}; +typedef struct registry_person REGISTRY_PERSON; + +// PERSON_URL +extern REGISTRY_PERSON_URL *registry_person_url_index_find(REGISTRY_PERSON *p, const char *url); +extern REGISTRY_PERSON_URL *registry_person_url_index_add(REGISTRY_PERSON *p, REGISTRY_PERSON_URL *pu) NEVERNULL WARNUNUSED; +extern REGISTRY_PERSON_URL *registry_person_url_index_del(REGISTRY_PERSON *p, REGISTRY_PERSON_URL *pu) WARNUNUSED; + +extern REGISTRY_PERSON_URL *registry_person_url_allocate(REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name, size_t namelen, time_t when); +extern REGISTRY_PERSON_URL *registry_person_url_reallocate(REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name, size_t namelen, time_t when, REGISTRY_PERSON_URL *pu); + +// PERSON +extern REGISTRY_PERSON *registry_person_find(const char *person_guid); +extern REGISTRY_PERSON *registry_person_allocate(const char *person_guid, time_t when); +extern REGISTRY_PERSON *registry_person_get(const char *person_guid, time_t when); +extern void registry_person_del(REGISTRY_PERSON *p); + +// LINKING PERSON -> PERSON_URL +extern REGISTRY_PERSON_URL *registry_person_link_to_url(REGISTRY_PERSON *p, REGISTRY_MACHINE *m, REGISTRY_URL *u, char *name, size_t namelen, time_t when); +extern void registry_person_unlink_from_url(REGISTRY_PERSON *p, REGISTRY_PERSON_URL *pu); + +#endif //NETDATA_REGISTRY_PERSON_H diff --git a/registry/registry_url.c b/registry/registry_url.c new file mode 100644 index 0000000..9ac3ce1 --- /dev/null +++ b/registry/registry_url.c @@ -0,0 +1,88 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#include "../daemon/common.h" +#include "registry_internals.h" + +// ---------------------------------------------------------------------------- +// REGISTRY_URL + +int registry_url_compare(void *a, void *b) { + if(((REGISTRY_URL *)a)->hash < ((REGISTRY_URL *)b)->hash) return -1; + else if(((REGISTRY_URL *)a)->hash > ((REGISTRY_URL *)b)->hash) return 1; + else return strcmp(((REGISTRY_URL *)a)->url, ((REGISTRY_URL *)b)->url); +} + +inline REGISTRY_URL *registry_url_index_add(REGISTRY_URL *u) { + return (REGISTRY_URL *)avl_insert(&(registry.registry_urls_root_index), (avl *)(u)); +} + +inline REGISTRY_URL *registry_url_index_del(REGISTRY_URL *u) { + return (REGISTRY_URL *)avl_remove(&(registry.registry_urls_root_index), (avl *)(u)); +} + +REGISTRY_URL *registry_url_get(const char *url, size_t urllen) { + // protection from too big URLs + if(urllen > registry.max_url_length) + urllen = registry.max_url_length; + + debug(D_REGISTRY, "Registry: registry_url_get('%s', %zu)", url, urllen); + + char buf[sizeof(REGISTRY_URL) + urllen]; // no need for +1, 1 is already in REGISTRY_URL + REGISTRY_URL *n = (REGISTRY_URL *)&buf[0]; + n->len = (uint16_t)urllen; + strncpyz(n->url, url, n->len); + n->hash = simple_hash(n->url); + + REGISTRY_URL *u = (REGISTRY_URL *)avl_search(&(registry.registry_urls_root_index), (avl *)n); + if(!u) { + debug(D_REGISTRY, "Registry: registry_url_get('%s', %zu): allocating %zu bytes", url, urllen, sizeof(REGISTRY_URL) + urllen); + u = callocz(1, sizeof(REGISTRY_URL) + urllen); // no need for +1, 1 is already in REGISTRY_URL + + // a simple strcpy() should do the job + // but I prefer to be safe, since the caller specified urllen + u->len = (uint16_t)urllen; + strncpyz(u->url, url, u->len); + u->links = 0; + u->hash = simple_hash(u->url); + + registry.urls_memory += sizeof(REGISTRY_URL) + urllen; // no need for +1, 1 is already in REGISTRY_URL + + debug(D_REGISTRY, "Registry: registry_url_get('%s'): indexing it", url); + n = registry_url_index_add(u); + if(n != u) { + error("INTERNAL ERROR: registry_url_get(): url '%s' already exists in the registry as '%s'", u->url, n->url); + freez(u); + u = n; + } + else + registry.urls_count++; + } + + return u; +} + +void registry_url_link(REGISTRY_URL *u) { + u->links++; + debug(D_REGISTRY, "Registry: registry_url_link('%s'): URL has now %u links", u->url, u->links); +} + +void registry_url_unlink(REGISTRY_URL *u) { + u->links--; + if(!u->links) { + debug(D_REGISTRY, "Registry: registry_url_unlink('%s'): No more links for this URL", u->url); + REGISTRY_URL *n = registry_url_index_del(u); + if(!n) { + error("INTERNAL ERROR: registry_url_unlink('%s'): cannot find url in index", u->url); + } + else { + if(n != u) { + error("INTERNAL ERROR: registry_url_unlink('%s'): deleted different url '%s'", u->url, n->url); + } + + registry.urls_memory -= sizeof(REGISTRY_URL) + n->len; // no need for +1, 1 is already in REGISTRY_URL + freez(n); + } + } + else + debug(D_REGISTRY, "Registry: registry_url_unlink('%s'): URL has %u links left", u->url, u->links); +} diff --git a/registry/registry_url.h b/registry/registry_url.h new file mode 100644 index 0000000..c684f1c --- /dev/null +++ b/registry/registry_url.h @@ -0,0 +1,35 @@ +// SPDX-License-Identifier: GPL-3.0-or-later + +#ifndef NETDATA_REGISTRY_URL_H +#define NETDATA_REGISTRY_URL_H 1 + +#include "registry_internals.h" + +// ---------------------------------------------------------------------------- +// URL structures +// Save memory by de-duplicating URLs +// so instead of storing URLs all over the place +// we store them here and we keep pointers elsewhere + +struct registry_url { + avl avl; + uint32_t hash; // the index hash + + uint32_t links; // the number of links to this URL - when none is left, we free it + + uint16_t len; // the length of the URL in bytes + char url[1]; // the URL - dynamically allocated to more size +}; +typedef struct registry_url REGISTRY_URL; + +// REGISTRY_URL INDEX +extern int registry_url_compare(void *a, void *b); +extern REGISTRY_URL *registry_url_index_del(REGISTRY_URL *u) WARNUNUSED; +extern REGISTRY_URL *registry_url_index_add(REGISTRY_URL *u) NEVERNULL WARNUNUSED; + +// REGISTRY_URL MANAGEMENT +extern REGISTRY_URL *registry_url_get(const char *url, size_t urllen) NEVERNULL; +extern void registry_url_link(REGISTRY_URL *u); +extern void registry_url_unlink(REGISTRY_URL *u); + +#endif //NETDATA_REGISTRY_URL_H |