#current beta regression

1 messages · Page 1 of 1 (latest)

modern sleet
ebon coveBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

kind harnessBOT
#

If you hit a regression on beta, treat it as a bug report, not something to workaround silently:

  1. First grab the exact version: openclaw --version and ideally openclaw status --all.
  2. Check whether there’s already an issue. For this one, #80322 looks like the right place.
  3. If you’re affected too, comment there with:
    • exact beta version
    • last known good version, if known
    • install method / OS
    • minimal repro steps
    • relevant redacted logs or openclaw status --all
    • whether it blocks you or is just annoying
  4. If it blocks real work, roll back/pin to the last known good or stable build until the beta is fixed.

For #80322 specifically, the issue already says the last known good was 2026.5.7 and the bad beta is 2026.5.10-beta.1, with suggested temporary workarounds being either stay on 2026.5.7 or move affected agents off the Codex runtime until upstream fixes the routing path.

Docs for diagnostics: https://docs.openclaw.ai/help/faq and https://docs.openclaw.ai/gateway/troubleshooting

modern sleet
#

I already wrote this issue, but I see ,that ut misses at least some marks like bug regression for some reason

kind harnessBOT
#

Yep — GitHub labels/marks like bug, regression, beta blocker are usually applied by maintainers/triage automation, not by the issue author.

What you can do:

  • Make sure the Bug type field says Regression (worked before, now fails) — yours does.
  • Make sure Beta release blocker says Yes — yours does.
  • Put the important signal in the title/body too, which you already did: Beta blocker + #80048 regression.
  • If you want to make it extra obvious, edit the first line of the summary to start with something like: Regression in 2026.5.10-beta.1; beta-blocking for SSH-backed Codex agents.

I wouldn’t open a duplicate issue just to get labels. The current issue has the right substance; labels can be added by maintainers/triage.

modern sleet
#

yep, is there any way I clould ensure that it is visible at least tto somebody responsible fro the release? it looks like such reports are a pretty rare case for openclaw - there only ten issues/PRs with such label

kind harnessBOT
#

Yes — the cleanest way is to make the GitHub issue self-triaging, without opening duplicates or pinging people.

I’d do this:

  1. Edit the title to make the release risk impossible to miss, e.g.
    Beta blocker regression: SSH-backed Codex agents lose exec/process routing in 2026.5.10-beta.1

  2. Add a short top comment like:
    Triage note: this appears to be a beta-blocking regression for users with Codex runtime + tools.exec.host="sandbox" + SSH sandbox backend. Last known good: 2026.5.7. First known bad: 2026.5.10-beta.1. No config workaround remains after #80048 except pinning/rolling back or moving off Codex runtime.

  3. Cross-link the regression-causing PR, which you already did. If there’s a beta release tracking issue/PR, add one calm comment there linking back to #80322. Don’t spam multiple places.

  4. If labels are missing, say explicitly in the issue body/comment: Suggested labels: bug, regression, beta blocker. Maintainers/triage can apply them.

That should be enough. A rare label is actually a good reason to be extra clear in the title/body, but the issue itself is the right place for this — not DMs/pings.

modern sleet
#

I am bit worried, that they will release this with that bug