Ceph » mgr

Clusters need to have debug_mgr 4/5 in order to see the useful mgr module INFO lines, e.g.:

2020-11-20T12:21:09.086+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] _maybe_adjust
2020-11-20T12:21:09.218+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] effective_target_ratio 0.0 0.55 0 6704959563235328.0
2020-11-20T12:21:09.218+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] Pool 'device_health_metrics' root_id -1 using 1.7097320113394475e-09 of space, bias 1.0, pg target 2.8096596053011588e-05 quantized to 1 (current 1)
2020-11-20T12:21:09.220+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] effective_target_ratio 0.0 0.55 0 6704959563235328.0
2020-11-20T12:21:09.221+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] Pool 'ecpool' root_id -2 using 0.0 of space, bias 1.0, pg target 0.0 quantized to 32 (current 64)
2020-11-20T12:21:09.223+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] effective_target_ratio 0.05 0.55 0 6704959563235328.0
2020-11-20T12:21:09.223+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] Pool 'cephfs_metadata' root_id -6 using 0.09090909090909091 of space, bias 4.0, pg target 5975.757575757576 quantized to 4096 (current 256)
2020-11-20T12:21:09.225+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] effective_target_ratio 0.5 0.55 0 6704959563235328.0
2020-11-20T12:21:09.225+0100 7f4f2ea3c700  0 [pg_autoscaler INFO root] Pool 'cephfs_data' root_id -2 using 0.9090909090909091 of space, bias 1.0, pg target 7469.696969696969 quantized to 8192 (current 8192)

...
2020-11-20T12:21:21.102+0100 7f4f31881700  0 [balancer INFO root] Optimize plan auto_2020-11-20_11:21:20
2020-11-20T12:21:21.102+0100 7f4f31881700  0 [balancer INFO root] Mode upmap, max misplaced 0.050000
2020-11-20T12:21:21.102+0100 7f4f31881700  0 [balancer INFO root] do_upmap
2020-11-20T12:21:21.102+0100 7f4f31881700  0 [balancer INFO root] pools ['cephfs_data']
2020-11-20T12:21:21.383+0100 7f4f31881700  0 [balancer INFO root] prepared 0/100 changes

These logs are not overly verbose, yet are super useful to help an operator understand what the mgr modules are doing.

I propose to change the default debug_mgr to 4/5 (from 1/5) or to change the mgr module "INFO" logs to be sent at ceph_level 1.