Further observations regarding the logging issue seen here:
The "cluster.daemon-log-level" is supposed to change the log level of the selfheal daemon.
While the "diagnostics.client-log-level" sets the client log level.
(It changes the fuse log level as well as the self heal log level.)
They show different behavior in situations.
One volume scenario:
1) When cluster.daemon-log-level is alone used, it works with a restart.
Change not reflected on fuse volfile but in the selfheal volfile.
2) When diagnostics.client-log-level is alone used. it makes the change on both the fuse and selfheal volfiles.
And is effective after a restart.
3) When cluster.daemon-log-level is used after the diagnostics.client-log-level, the level set by diagnostics.client
becomes effective and not the cluster.daemon. This is the reason we see that the log level changes are not becoming effective.
4) When tried in the other direction (cluster.daemon-log-level and then the diagnostics.client-log-level), then
diagnostics.client-log-level becomes effective. Note: this changes both the self heal and the fuse log level.
5) When the diagnostics level is INFO, then changing the cluster.daemon level works.
While diagnostics was debug, it is not possible to set the cluster daemon to info.
TWO volumes scenario:
6) Both log level works fine. The diagnostics log level changes only for the fuse(on the particular volume where it was set) and not for the selfheal.
Self heal follows the cluster daemon log level.
If the diagnostics level is supposed to take precedence over the daemon log level, then the situation 5 should be allowed.
If that happens the way the 2 volume setup works is messed up.
This is the confusing part I have seen regarding the log level.