Bug #55623: cephadm: call handling causing process to not exit when container does - Orchestrator - Ceph

Actions

Copy link

Bug #55623

closed

cephadm: call handling causing process to not exit when container does

Added by Adam King almost 2 years ago. Updated almost 2 years ago.

Status:

Resolved

Priority:

High

Assignee:

Adam King

Category:

Target version:

% Done:

Source:

Tags:

Backport:

quincy

Regression:

Yes

Severity:

2 - major

Reviewed:

Affected Versions:

ceph-qa-suite:

Pull request ID:

46245

Crash signature (v1):

Crash signature (v2):

Description

Basically, in certain environments with certain commands (specifically saw on a host with 102 multipath devices, running a `cephadm ceph-volume inventory` command), when cephadm runs a subprocess like `podman run . . .` we're seeing situations where the command in the container will complete, the container will exit but the actual process for the `podman run` command does not. This causes cephadm to hang indefinitely waiting for the process to finish.

This seems to be caused by a recent change to the `call` function in the cephadm binary as reverting the change locally caused the issue to go away.

Related issues 1 (0 open — 1 closed)