Resetting
How to make a Stroom component start again from a known state, and what each reset costs you.
Sometimes the fix is to make something start again. A pipeline was wrong and the output it produced is wrong with it; an index holds data in a shape you have since changed; a cache is serving a document you have already corrected.
Each page here covers one component. They share a shape, because the same three questions apply every time:
- What does it actually clear? Rarely only the thing you had in mind.
- What is lost, and can it be rebuilt? Most of Stroom's derived data can be rebuilt from the stream store, which is the strongest argument for keeping raw data. Some things cannot.
- What has to happen afterwards? A reset usually leaves work to redo, and a system that has been reset but not reprocessed looks like a system that has lost data.
Warning
Read the whole page before running anything on it. Several of these operations are irreversible, and two of them delete data that no amount of reprocessing will bring back.
Which one do you want?
| Symptom | Page |
|---|---|
| The output is wrong, and the pipeline is now right | Reprocess data |
| New data processes but old data was skipped | Reset a processor filter |
| Searches are wrong, or you changed an index field type | Reset an index |
| A corrected document is still behaving as it did before | Clear caches |
| Reference lookups return stale or deleted values | Purge reference data |
| A Plan B store is full, stale, or has the wrong schema | Reset a Plan B store |
| Nobody can sign in as admin | Reset a user's password |