Skip to content

builder: longer timeouts and more retries for APT to avoid aborted builds due to network issues #9936

@adrelanos

Description

@adrelanos

Qubes OS release

R4.2

Brief summary

Quote @marmarek in QubesOS/updates-status#5605 (comment)

Maybe there should be some (better) retry mechanism? Documentation suggests Acquire::Retries setting defaults to 3, and indeed I see 3 attempts in the log. But maybe it's not enough? Or maybe it fails due to the current tor circuit being broken and retry should use a different one to have a chance (not sure how to do that, or if that's even possible)? Or maybe when it fails on name resolution (log has Something wicked happened resolving 'deb.qubes-os.org:https' (-4 - Non-recoverable failure in name resolution).) it doesn't really retry? I'm quite confused here, because it stops working in the middle of template build, for example installing qubes-core-agent-dom0-updates some lines above worked. And also installing a lot of Debian and Whonix packages earlier worked too...

Steps to reproduce

Build a Debian based Qubes Template.

Expected behavior

Long timeouts and many retries.

Actual behavior

Bails out due to network issues.

Metadata

Metadata

Assignees

No one assigned

    Labels

    C: WhonixThis issue pertains to Whonix templates or standalones.C: builderThis issue pertains to the Qubes Builder.P: defaultPriority: default. Default priority for new issues, to be replaced given sufficient information.affects-4.2This issue affects Qubes OS 4.2.community templateThis issue pertains to a community-maintained template.needs diagnosisRequires technical diagnosis from developer. Replace with "diagnosed" or remove if otherwise closed.

    Type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions