fix: better query manager and allow setting cookie for dev forward#19692
fix: better query manager and allow setting cookie for dev forward#19692vogievetsky wants to merge 2 commits into
Conversation
FrankChen021
left a comment
FrankChen021
left a comment
There was a problem hiding this comment.
| Severity | Findings |
|---|---|
| P0 | 0 |
| P1 | 0 |
| P2 | 2 |
| P3 | 0 |
| Total | 2 |
| Severity | Findings |
|---|---|
| P0 | 0 |
| P1 | 0 |
| P2 | 2 |
| P3 | 0 |
| Total | 2 |
Reviewed 6 of 6 changed files.
Found two compaction-details integration issues: the Overlord-only fallback does not identify autocompact rows, and the formatter receives the supervisor wrapper instead of its nested compaction config.
This is an automated review by Codex GPT-5.6-Sol
| aggregateLag, | ||
| )}`}</span> | ||
| ) : null; | ||
| } else if (original.type === 'autocompact') { |
There was a problem hiding this comment.
[P2] Handle autocompact type in the non-SQL fallback
On the non-SQL path above, each row's type is derived exclusively from sup.spec.tuningConfig.type. A CompactionSupervisorSpec serializes its type at sup.spec.type and its config at sup.spec.spec, so autocompact rows have original.type === undefined; this new branch therefore never executes when the console is pointed directly at an Overlord. Include sup.spec.type when mapping fallback rows so Details works in this supported mode.
There was a problem hiding this comment.
Good find. This is a but in the console in general (not introduced in this PR). Fixed it and will now make the PR as a bug fix.
| if (!supervisorStatusPayload || !original.spec) return null; | ||
| return formatCompactionInfo({ | ||
| status: supervisorStatusPayload, | ||
| config: original.spec as CompactionConfig, |
There was a problem hiding this comment.
[P2] Pass the nested compaction config
original.spec is the serialized supervisor wrapper ({ type: "autocompact", spec: DataSourceCompactionConfig, ... }), not the compaction config itself. Passing the wrapper makes formatCompactionInfo read skipOffsetFromLatest and inputSegmentSizeBytes from the wrong level, so fully compacted statuses with skipped segments always claim the default P1D exclusion and omit any legacy segment-size exclusion. Pass original.spec.spec instead.
There was a problem hiding this comment.
wow. Good spot.
FrankChen021
left a comment
FrankChen021
left a comment
There was a problem hiding this comment.
Both compaction-details findings are addressed: the Overlord fallback now recognizes spec.type, and Details passes the nested spec.spec compaction config to the formatter. No additional correctness issues found.
Reviewed 6 of 6 changed files.
This is an automated review by Codex GPT-5.6-Sol
2 participants
Make it so that background refreshes do not re-jigger the table so much
Also: