Project

General

Profile

Bug #21024

Perfcounters: wrong latency information from admin socket

Added by Aleksei Zakharov over 6 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
OSD
Target version:
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
fs
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

Hello.
We found out that for some OSD's op_w_latency and op_r_latency counters have values of 0:

ceph --admin-daemon /var/run/ceph/ceph-osd.66.asok perf dump | grep -A3 op_r_lat
        "op_r_latency": {
            "avgcount": 0,
            "sum": 0.000000000
        },

We use collectd ceph plugin and it returns no value if 0 is present by socket. We tried to reset counters with `perf reset all`, but it didn't help.
These are production serves, so we didn't try to restart OSD's yet. This is not a single case, we have more then one OSD with wrong perf counters.
Hope for help.

Also available in: Atom PDF