Project

General

Profile

Bug #1845

"recovery_ops" performance counter isn't decreased

Added by Szymon Szypulski almost 12 years ago. Updated almost 12 years ago.

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

0%

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

Description

I'm generating osd statistics based on performance sockets like described here - http://ceph.newdream.net/wiki/Perfomance_counters. I've two OSD daemons. I'm requesting "recovery_ops", first shows 0 all time and second shows 7 also all time.

I've contacted with gregaf through irc and he suggested that this counter doesn't work as intended

19:46 [ gregaf1] chaos_: hmm, looks like whoever implemented the counter didn't finish — it never decrements!

Effect:

History

#1 Updated by Sage Weil almost 12 years ago

The counters are counting events and never decrement. Normally collectd will divide the change by time to give you something interesting... are you using collectd, or something else?

#2 Updated by Szymon Szypulski almost 12 years ago

I'm using munin, wrote plugin myself and I don't divide this value by anything. If it shouldn't decrement please close this as invalid, but it think it should show present status of osd daemon.

#3 Updated by Sage Weil almost 12 years ago

  • Status changed from New to Rejected

Right, it should never decrement. Closing!

Also available in: Atom PDF