26 Sep
2019
26 Sep
'19
5:24 a.m.
Hi Miha,
interesting observation, I don't think we've noticed this before. Would
you mind submitting a bug report about this on our tracker, including
these logs?
https://tracker.ceph.com/projects/mgr/issues/new
Thanks in advance!
Lenz
On 9/26/19 10:01 AM, Miha Verlic wrote:
On 24. 09. 19 14:53, Lenz Grimmer wrote:
--
SUSE Software Solutions Germany GmbH - Maxfeldstr. 5 - 90409 Nuernberg
GF: Felix Imendörffer, HRB 247165 (AG Nürnberg)
On 9/24/19 1:37 PM, Miha Verlic wrote:
There is systemd-tmpfiles-clean.timer, but it ignores PrivateTmp folders.
But as said, crt and key files do exist in tmp and according to
timestamps they were created when ceph-mgr daemon was started; it seems
to me like ceph-mgr simply starts looking for wrong filenames after a while.
Going through ceph-mgr logs I'd say the problem is internal webserver is
restarted, but ceph-mgr is not notified about new crt/key files:
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
Bus STOPPING
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
HTTP Server cherrypy._cpwsgi_server.CPWSGIServer(('::', 8443)) shut down
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
Stopped thread '_TimeoutMonitor'.
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
Bus STOPPED
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
Bus STARTING
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
Started monitor thread '_TimeoutMonitor'.
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: 2019-09-19 12:54:16.618
7f3a796e4700 -1 client.0 error registering admin socket command: (17)
File exists
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: 2019-09-19 12:54:16.618
7f3a796e4700 -1 client.0 error registering admin socket command: (17)
File exists
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: 2019-09-19 12:54:16.618
7f3a796e4700 -1 client.0 error registering admin socket command: (17)
File exists
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: 2019-09-19 12:54:16.618
7f3a796e4700 -1 client.0 error registering admin socket command: (17)
File exists
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: 2019-09-19 12:54:16.618
7f3a796e4700 -1 client.0 error registering admin socket command: (17)
File exists
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
Serving on :::8443
Sep 19 12:54:16 cephtest01 ceph-mgr[2247]: [19/Sep/2019:12:54:16] ENGINE
Bus STARTED I've got slightly different problem. After a
few days of running fine,
dashboard stops working because it is apparently seeking for wrong
certificate file in /tmp. If I restart ceph-mgr it starts to work again.
Does the restart trigger the creation of a similar-looking file in /tmp?
I wonder if there's some kind of cron job that cleans up the /tmp
directory every now and then...