summaryrefslogtreecommitdiffstats
path: root/doc/mgr/crash.rst
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.