Woodpecker daily cron builds get stuck

[martint17r]

Component

agent

Describe the bug

After about two days our @daily cron builds get stuck and error immediately or timeout after 60 minutes:

• fatal: could not read Username for 'https://git...de': No such device or address\nexit status 128
• Error response from daemon: unable to find network with name or ID wp_01hpbtc4ccn66zzwk22mvd33a0_0_default: network not found

Restarting the agent resolves the problem and the @daily cron jobs work for a few days.

What actions would you recommend me to take to debug this problem?

System Info

{"source":"https://github.com/woodpecker-ci/woodpecker","version":"2.3.0"} - also occured with 2.2.0
Debian bookworm
podman 4.3.1
podman server and agent processes are managed through systemd units created by run podman generate.
gitea 1.21.5
WOODPECKER_AUTHENTICATE_PUBLIC_REPOS=true
MariaDB 10.11.6

Additional context

No response

Validations

• Read the docs.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• Checked that the bug isn't fixed in the next version already [https://woodpecker-ci.org/faq#which-version-of-woodpecker-should-i-use]

[martint17r]

This problem has not occured after we upgraded to 2.4.1

[martint17r]

The problem reoccured with 2.6.0 - now the error is "pipeline definition not found".