Description of problem:

rex pull provider job using rhel9 backend on capsule consistently fails in automation, with ssh2.exceptions.Timeout.

Trying out with manual setup I got two kinds of errors. On an initial rex run I got failed job with

Unexpected event #<Actions::ProxyAction::ProxyActionStopped: Actions::ProxyAction::ProxyActionStopped>

On a subsequent run, the job remains in pending state with the following msg reeling:

Error loading data from Capsule: NoMethodError - undefined method `each_with_index' for nil:NilClass

which probably accounts for the timeouts whe see in automation

How reproducible:

Always

Is this issue a regression from an earlier version:

We didn't have rhel9 backend in earlier versions so technically no.

Steps to Reproduce:

1. have a setup with external mqtt capsule on rhel9 backend and client with yggdrasil client.

2. run a rex job

Actual behavior:
Issues described above

Expected behavior:
Successful rex run

Business Impact / Additional info:

• Setup available
• on rhel8 backend the same tests consistently pass
• interestingy, there is one rex pull test that passes on rhel9 – it runs the job against client with stopped ygdrassil, then starts it while the job is pending. In this scenario the job succeeds.