You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On a1 the deps agent runs with sandbox.runtime: job. Clicking 📄 full log or 📋 log text on its card returns 404; in-pod GET /api/agents/deps/log answers capturing pane for deps: exit status 1. The same buttons work for tmux agents (scanner).
Root cause. Every agent gets a tmuxSession name at construction, so CaptureFullLog runs tmux capture-pane for a sandboxed agent that has no session; the handler maps every capture error to 404. The run's output is never kept anywhere durable: the kubejob launcher collects the pod log into memory and deletes the Job on success (failed Jobs expire at ttl_seconds, 1h), and the workspace transcript .hive/sandbox-transcript.log lives under the sandbox root that sweepWorkspaces removes after 2h (workspace cleanup: removed stale dir agent=sandbox dir=deps-1793182030794632340). Result: the owner wanted the raw agent output to tune the kick template and it was unrecoverable.
Expected. For sandboxed agents hive should persist the captured Job/container stdout+stderr per kick (alongside the SandboxRun from #14845) and serve that from the full-log endpoint and the kick history; the card buttons should point at what exists, and before the first run say so instead of 404.
@clubanderson this issue was closed by @hivecommons-hive[bot] as completed.
If this was closed by a merged PR, the fix only takes effect once it ships in a tagged release and your deployment updates to it — check your running version against the fix before concluding the fix failed.
If you are still seeing the problem, comment /reopen and it will be reopened for you — you do not need write access. Please add what you are still observing (running version, logs, steps) so it is not closed the same way again.
Activity
github-actions commentedon Oct 28, 2026
@clubanderson this issue was closed by @hivecommons-hive[bot] as completed.
If this was closed by a merged PR, the fix only takes effect once it ships in a tagged release and your deployment updates to it — check your running version against the fix before concluding the fix failed.
If you are still seeing the problem, comment
/reopenand it will be reopened for you — you do not need write access. Please add what you are still observing (running version, logs, steps) so it is not closed the same way again.