Bug #42894: kclient: if there has at least one MDS still not laggy the mount will fail - CephFS - Ceph

Actions

Copy link

Bug #42894

closed

kclient: if there has at least one MDS still not laggy the mount will fail

Added by Xiubo Li over 4 years ago. Updated about 4 years ago.

Status:

Resolved

Priority:

Normal

Assignee:

Xiubo Li

Category:

Target version:

% Done:

Source:

Development

Tags:

Backport:

Regression:

Severity:

3 - minor

Reviewed:

Affected Versions:

ceph-qa-suite:

Component(FS):

kceph

Labels (FS):

Pull request ID:

Crash signature (v1):

Crash signature (v2):

Description

In case:

ceph fs dump
[...]

max_mds 3
in 0,1,2
up {0=5139,1=4837,2=4985}
failed
damaged
stopped
data_pools [2]
metadata_pool 1
inline_data disabled
balancer
standby_count_wanted 1
[mds.a{0:5139} state up:active seq 7 laggy since 2019-11-20T01:04:13.040701-0500 addr v1:192.168.195.165:6813/2514516359]
[mds.b{1:4837} state up:active seq 6 addr v1:192.168.195.165:6815/1921459709]
[mds.f{2:4985} state up:active seq 6 laggy since 2019-11-20T01:04:13.040685-0500 addr v1:192.168.195.165:6814/3730607184]

The m->m_num_laggy == 2, but there still has one MDS in (up:active & !laggy) state. In this case if the mount request choose the mds.a, there still has the IO errors and failure. A better choice is that it can choose the mds.b instead. Currently the ceph_mdsmap_is_cluster_available() will just return false if there has any MDS is laggy.