Actions
CDM 01-MAR-2023 » History » Revision 6
« Previous |
Revision 6/8
(diff)
| Next »
Prashant D, 03/02/2023 12:48 AM
CDM 01-MAR-2023¶
- Better handling of cluster log messages from monstore perspective
Problem :
- every clog message goes to mon db
- debug/warning generating clog messages 1000s of times per seconds is not useful
- In catastrophic failure situation, mon db growing exponentially
Need of storing logm entries in mon db
- for "ceph log last" to get historical clog entries
- ceph dashboard (mgr is subscriber of log-info which propogate clog to dashboard module)
Solution:
- Stop writing logm entries to mon db if there is an excessive entries getting generated
- filter out clog DBG entries
Updated by Prashant D about 1 year ago · 6 revisions