Project

General

Profile

Actions

Feature #23374

closed

Get test suite passing on Debian 13

Added by Brett Smith 3 months ago. Updated about 2 months ago.

Status:
Resolved
Priority:
Normal
Assigned To:
Category:
Tests
Target version:
Story points:
-
Release relationship:
Auto

Description

There are some minor failures caused by library drift, etc. See attached log of a recent run.

  • sdk/cli - curb install failure, investigating
  • sdk/cwl - See #22695, I tried a fix but I think more work is required
  • lib/config - We should just drop support for DSA keys since OpenSSH has too
  • lib/controller/localdb - Test regexp needs update to accommodate minor formatting change
  • lib/crunchrun - This is a configuration problem on my VM, not real
  • services/keep-web - Needs investigation
  • tools/keep-rsync - Also needs investigation, looks potentially related
  • services/workbench2_units - I think this is an error in how I set up my temp dir, probably not real

I may split individual failures out into dedicated tickets if they become involved enough. This ticket can make a branch for miscellaneous small fixes.


Files

debian13.log.xz (113 KB) debian13.log.xz Brett Smith, 12/12/2025 02:24 PM
23374retest.log.xz (26.7 KB) 23374retest.log.xz Brett Smith, 12/13/2025 06:44 PM
23374test.log.xz (110 KB) 23374test.log.xz Brett Smith, 12/13/2025 06:44 PM

Subtasks 1 (0 open1 closed)

Task #23378: Review 23374-trixie-testsResolvedBrett Smith12/16/2025Actions

Related issues 1 (1 open0 closed)

Related to Arvados Epics - Idea #23170: Add support for Debian 13 (trixie)NewActions
Actions #1

Updated by Brett Smith 3 months ago

  • Related to Idea #23170: Add support for Debian 13 (trixie) added
Actions #2

Updated by Brett Smith 3 months ago

Actions #3

Updated by Brett Smith 3 months ago

Brett Smith wrote:

  • sdk/cli - curb install failure, investigating
  • tools/keep-rsync - Also needs investigation, looks potentially related
  • services/workbench2_units - I think this is an error in how I set up my temp dir, probably not real

All of these seem to be local setup/process bugs.

  • lib/controller/localdb - Test regexp needs update to accommodate minor formatting change
  • services/keep-web - Needs investigation

Note both of these involve cadaver. It seems like the version in Debian 13 makes substantial changes. Still working on the second one.

Actions #4

Updated by Brett Smith 3 months ago

https://notroj.github.io/cadaver/

We are upgrading from 0.24 (Debian 12) to 0.26 (Debian 13). I note these changes especially:

  • Rewrite of ‘copy’ and ‘move’ commands:
    • a source collection is moved/copied inside a destination coll rather than overwriting it.
    • more error cases are disallowed (e.g. ‘mv / /foo’)
    • a source collection with a non-collection destination now fails

Updated by Brett Smith 3 months ago

23374-trixie-tests @ 46f06ad397a5e1c978d69b552d3294619401604e - developer-run-tests: #4978

Updates various tests so they pass on Debian 13. Each individual commit has a rationale. The only change that took any real work was updating the keep-web tests to use the new cadaver rename command.

  • All agreed upon points are implemented / addressed. Describe changes from pre-implementation design.
    • Yes
  • Anything not implemented (discovered or discussed during work) has a follow-up story.
    • #22695 I guess. That does ultimately block Debian 13 support, but note it does not block upgrading Jenkins worker nodes since they run older Python built from source.
  • Code is tested and passing, both automated and manual, what manual testing was done is described.
    • See attached log of a test run on Debian 13, including one retry. Reasons original tests failed:
      • sdk/cli: Not sufficiently isolated from settings.conf. Rerun removed it. I could file a bug but realistically the solution will be #22390.
      • services/fuse: No idea. I have not seen this before or since. Cleared with just a reboot.
      • lib/controller/localdb: My development environment runs with umask 027. The fix was chmod -R o+rX lib/controller/localdb/login_docker_test
      • lib/crunchrun: The cgroup failures are a limitation of my development VM, it doesn't have all the cgroup stuff passed through. I'm guessing the original FUSE failures happened for the same reason as the FUSE tests.
  • Tested code incorporates recent main branch changes.
    • Yes
  • New or changed UI/UX and has gotten feedback from stakeholders.
    • N/A
  • Documentation has been updated.
    • N/A
  • Behaves appropriately at the intended scale (describe intended scale).
    • No change
  • Considered backwards and forwards compatibility issues between client and server.
    • N/A
  • Follows our coding standards and GUI style guidelines.
    • Yes
Actions #6

Updated by Brett Smith 3 months ago

  • Subtask #23378 added
Actions #7

Updated by Tom Clegg 3 months ago

  • Status changed from New to In Progress
Actions #8

Updated by Tom Clegg 3 months ago

nit 1: newCadaverOptions could be more concise if we import x/mod/semver and do

match := regexp.MustCompile(`(?m)^cadaver\s+(\d+\.\d+)\b`).FindSubmatch(stdout)
// ...
if semver.Compare("v"+match[1], "v0.26") < 0 {
        // ...

nit 2: we could make it newCadaverOpts(c *check.C) and call c.Assert() instead of returning errors, to get slightly more helpful gocheck output.

Either way, LGTM.

Actions #9

Updated by Brett Smith 3 months ago

Tom Clegg wrote in #note-8:

Either way, LGTM.

Thanks, both taken at ca41437f817dfb5651f3486108724a1b45fb43ac, developer-run-tests: #4981 - I'll merge tomorrow as long as tests pass and there are no objections.

Actions #10

Updated by Tom Clegg 3 months ago

LGTM, thanks.

Actions #11

Updated by Brett Smith 3 months ago

  • Status changed from In Progress to Resolved
Actions #13

Updated by Brett Smith about 2 months ago

  • Release set to 84
Actions

Also available in: Atom PDF