blob: 656c7a2d4ef31c5d59d5b056170ea661dd92fcab (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
|
Crash Module
============
The crash module collects information about daemon crashdumps and stores
it in the Ceph cluster for later analysis.
Enabling
--------
The *crash* module is enabled with::
ceph mgr module enable crash
The *crash* upload key is generated with::
ceph auth get-or-create client.crash mon 'profile crash' mgr 'profile crash'
On each node, you should store this key in
``/etc/ceph/ceph.client.crash.keyring``.
Automated collection
--------------------
Daemon crashdumps are dumped in ``/var/lib/ceph/crash`` by default; this can
be configured with the option 'crash dir'. Crash directories are named by
time and date and a randomly-generated UUID, and contain a metadata file
'meta' and a recent log file, with a "crash_id" that is the same.
These crashes can be automatically submitted and persisted in the monitors'
storage by using ``ceph-crash.service``.
It watches the crashdump directory and uploads them with ``ceph crash post``.
``ceph-crash`` tries some authentication names: ``client.crash.$hostname``,
``client.crash`` and ``client.admin``.
In order to successfully upload with ``ceph crash post``, these need
the suitable permissions: ``mon profile crash`` and ``mgr profile crash``
and a keyring needs to be in ``/etc/ceph``.
Commands
--------
::
ceph crash post -i <metafile>
Save a crash dump. The metadata file is a JSON blob stored in the crash
dir as ``meta``. As usual, the ceph command can be invoked with ``-i -``,
and will read from stdin.
::
ceph crash rm <crashid>
Remove a specific crash dump.
::
ceph crash ls
List the timestamp/uuid crashids for all new and archived crash info.
::
ceph crash ls-new
List the timestamp/uuid crashids for all newcrash info.
::
ceph crash stat
Show a summary of saved crash info grouped by age.
::
ceph crash info <crashid>
Show all details of a saved crash.
::
ceph crash prune <keep>
Remove saved crashes older than 'keep' days. <keep> must be an integer.
::
ceph crash archive <crashid>
Archive a crash report so that it is no longer considered for the ``RECENT_CRASH`` health check and does not appear in the ``crash ls-new`` output (it will still appear in the ``crash ls`` output).
::
ceph crash archive-all
Archive all new crash reports.
Options
-------
* ``mgr/crash/warn_recent_interval`` [default: 2 weeks] controls what constitutes "recent" for the purposes of raising the ``RECENT_CRASH`` health warning.
* ``mgr/crash/retain_interval`` [default: 1 year] controls how long crash reports are retained by the cluster before they are automatically purged.
|
