Hi, We are running 10 servers with glusterfs on it. To form 1 big file systeem. This works great, but I have only 1 issue, Logs. We use a central syslog server, which recieves all logs of our servers. This is nice to view.manage and monitor logs. Only for glusterfs I have to log in ever servers, to view logs, and I have to run on every server a nagios plugin, to watches the logs of glusterfs. It would be really nice, if glusterfs can log all messages to syslog. So I can watch my logs and monitor them from 1 place. I know, I can make a script or use a external program, to parse the logs and send the logs to syslog. But I think this function should be in glusterfs. Grtz, Jiri
can you try 'gluster volume set <VOL> diagnostics.brick-sys-log-level WARNING" ?
@Amar, thnx this works. I guess, when I set this 'gluster volume set <VOL> diagnostics.brick-sys-log-level WARNING" I get the warnings, errors and critical messages? But what about the info messages?
(In reply to comment #2) > @Amar, thnx this works. > Thanks for confirming. > I guess, when I set this 'gluster volume set <VOL> > diagnostics.brick-sys-log-level WARNING" I get the warnings, errors and > critical messages? Yes, thats right. > > But what about the info messages? You can give 'INFO' instead of 'WARNING' in the above message. Please check the 'gluster volume set help' :-) Let us know if this info works for you, can we go ahead and close the bug?
Nop, I can not set INFO. root@gluster-brick-01n1:~# gluster volume set glusterfsvol02 diagnostics.client-sys-log-level INFO option sys-log-level INFO: 'INFO' is not valid (possible options are WARNING, ERROR, CRITICAL.) Set volume unsuccessful btw, I'm using glusterfs 3.3.1
CHANGE: http://review.gluster.org/4499 (Debug/io-stats: Enable INFO level in syslog) merged in master by Vijay Bellur (vbellur)
Create to see, that INFO is added :) But after seeing the code, I have 1 question. Why not include every message type? for syslog .value = { "WARNING", "ERROR", "INFO", "CRITICAL"} for brick-log-level .value = { "DEBUG", "WARNING", "ERROR", "INFO","CRITICAL", "NONE", "TRACE"}
DEBUG, TRACE levels are very much a glusterfs debug levels, and *should* not be used in a 'admin' setup. That is the reason for not having these options for syslog.
Ok, that is clear. Thanks for explaning it.