Observed: Repo failure results in defunct processes

[greenaar]

When the target repo is not available (connection failure, maintenance, etc), the git-remote-http process associated seems to hang around

nobody 10813 1.0 0.2 56008 45072 ? Ssl Oct21 7:58 | _ /usr/bin/katafygio --local-dir=/var/lib/katafygio/data --healthcheck-port=8080 --git-url=https://REDACTED:PASSWORD@TARGETSERVER/scm/tool/kube
nobody 24516 0.0 0.0 0 0 ? Z 01:48 0:00 | _ [git-remote-http]
nobody 24810 0.0 0.0 0 0 ? Z 01:48 0:00 | _ [git-remote-http]
nobody 24999 0.0 0.0 0 0 ? Z 01:48 0:00 | _ [git-remote-http]
nobody 25137 0.0 0.0 0 0 ? Z 01:48 0:00 | _ [git-remote-http]
nobody 25288 0.0 0.0 0 0 ? Z 01:48 0:00 | _ [git-remote-http]
nobody 25435 0.0 0.0 0 0 ? Z 01:49 0:00 | _ [git-remote-http]
nobody 25538 0.0 0.0 0 0 ? Z 01:49 0:00 | _ [git-remote-http]
nobody 25658 0.0 0.0 0 0 ? Z 01:49 0:00 | _ [git-remote-http]
nobody 25776 0.0 0.0 0 0 ? Z 01:49 0:00 | _ [git-remote-http]
nobody 25975 0.0 0.0 0 0 ? Z 01:49 0:00 | _ [git-remote-http]

Killing the pod wipes these out, but that's by no means idea.

Are there any easy solutions?

[bpineau]

Hi @greenaar , thanks for reporting that.
Good catch! We should run katafygio with tini, so ssh helpers would get properly reaped.

[bpineau]

Hey @greenaar , this should be resolved by the last release (0.7.3). Thanks for the report!