Bug #62638: mgr/cephadm is not defining haproxy tcp healthchecks for Ganesha - Orchestrator - Ceph

Actions

Copy link

Bug #62638

closed

mgr/cephadm is not defining haproxy tcp healthchecks for Ganesha

Added by Paul Cuzner 8 months ago. Updated about 1 month ago.

Status:

Resolved

Priority:

High

Assignee:

Avan Thakkar

Category:

cephadm/services

Target version:

% Done:

Source:

Tags:

backport_processed

Backport:

squid, reef

Regression:

Severity:

3 - minor

Reviewed:

Affected Versions:

Ceph - v18.1.0, Ceph - v18.1.1, Ceph - v18.1.2, Ceph - v18.1.3, Ceph - v18.2.0, Ceph - v18.2.1

ceph-qa-suite:

Pull request ID:

53840

Crash signature (v1):

Crash signature (v2):

Description

Normally, I'd expect the backend server definition for haproxy to include healthchecks to take servers out of 'rotation' when they are down. However, looking at the Reef, the backend definition doesn't include a check
i.e.

backend backend
  mode        tcp
  balance     source
  hash-type   consistent
  server nfs.ganesha.0 10.1.37.143:12049
  server nfs.ganesha.1 10.1.39.54:12049

Resolving this could be as simple as adding check to the server definitions.

Related issues 2 (0 open — 2 closed)