Bug #20886: Bad perf counters in ceph-mgr for MDS - mgr - Ceph

Actions

Copy link

Bug #20886

closed

Bad perf counters in ceph-mgr for MDS

Added by John Spray over 6 years ago. Updated over 6 years ago.

Status:

Can't reproduce

Priority:

Normal

Assignee:

John Spray

Category:

Target version:

% Done:

Source:

Tags:

Backport:

Regression:

Severity:

3 - minor

Reviewed:

Affected Versions:

ceph-qa-suite:

Pull request ID:

Crash signature (v1):

Crash signature (v2):

Description

Could be a graphing bug but needs narrowing down -- looking at the FS page on the dashboard for the lab cluster (running 12.1.2) is showing whacky stuff.

The mds.inodes count is orders of magnitude out (although trending in the correct way), and the mds_server.handle_client_request rate is showing unreasonably large values in periodic peaks (the real behaviour according to daemonperf is a much steadier tens-hundreds of requests per second).

Files

Screenshot from 2017-08-02 19-10-52.png (43.2 KB) Screenshot from 2017-08-02 19-10-52.png

John Spray, 08/02/2017 06:11 PM