Fix PDisk shred free chunks in transit and after vdisk deletion #18279
Conversation
|
🟢 |
the-ancient-1
changed the title
the-ancient-1
changed the title
|
⚪ DetailsTest history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ DetailsTest history | Ya make output | Test bloat
⚪ DetailsTest history | Ya make output | Test bloat | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
rosik
changed the title
There was a problem hiding this comment.
Pull Request Overview
This PR fixes issues with PDisk shredding by ensuring that chunks flagged for deletion are correctly marked as dirty and processed appropriately during state transitions.
- Marks chunks as dirty during owner kill when shred mode is enabled.
- Adds a loop to check and schedule shredding for chunks in transition between states.
| << " state# " << state.ToString() | ||
| << ", there are already ContinueShredsInFlight# " << ContinueShredsInFlight.load() | ||
| << " so just wait for it to arrive. " | ||
| << " ShredGeneration# " << ShredGeneration); |
There was a problem hiding this comment.
The early return within the loop means that only the first matching dirty chunk is processed per call. If this is intended, please add a comment to clarify that scheduling only one continue shred per invocation is by design.
| << " ShredGeneration# " << ShredGeneration); | |
| << " ShredGeneration# " << ShredGeneration); | |
| // Early return is intentional: only the first matching dirty chunk is processed per invocation. | |
| // This ensures that scheduling only one continue shred per invocation is by design. |
| << ", scheduling ContinueShred. " | ||
| << " ShredGeneration# " << ShredGeneration); | ||
| ContinueShredsInFlight++; | ||
| PCtx->ActorSystem->Schedule(TDuration::MilliSeconds(50), |
There was a problem hiding this comment.
[nitpick] The literal delay value '50' in TDuration::MilliSeconds(50) could be replaced with a named constant to improve readability and maintainability.
| PCtx->ActorSystem->Schedule(TDuration::MilliSeconds(50), | |
| PCtx->ActorSystem->Schedule(TDuration::MilliSeconds(CONTINUE_SHRED_DELAY_MS), |
2 participants
Changelog entry
...
Changelog category
Description for reviewers
Close #18217
Close #18218