Depends on D3649
Adding a terminate callback so we wait until the chunks upload is finished (if it ever should be performed in a current case).
Details
Details
cd services yarn run-backup-service
Diff Detail
Diff Detail
- Repository
- rCOMM Comm
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
| services/backup/docker-server/contents/server/src/Reactors/server/SendLogReactor.h | ||
|---|---|---|
| 165–169 ↗ | (On Diff #11174) | I know we're going to search for a better alternative, but wanted to emphasize that this solution is fragile. We don't have any guarantees about when the scheduled write will finish. Is this possible to have a callback added to scheduleSendingDataChunk which will be called when the operation is finished? We could call storeInDatabase in this callback. |
| services/backup/docker-server/contents/server/src/Reactors/server/SendLogReactor.h | ||
|---|---|---|
| 165–169 ↗ | (On Diff #11174) | Hmm, could you please throw some light on what exactly will be different? Here we're waiting with the CV, you're suggesting a callback. These are two solutions to the same problem and I don't really see a difference in the outcome. In other words, how would using a callback make a difference for this:
? |
| services/backup/docker-server/contents/server/src/Reactors/server/SendLogReactor.h | ||
|---|---|---|
| 165–169 ↗ | (On Diff #11174) | Overall, writing the code that uses synchronization is a lot more complicated - there are multiple unintuitive mistakes that can easily be made without deep understanding. On the other hand, using a callback is intuitive. It also helps with readability - we're explicit about dependencies. But overall, if this callback need a synchronization, then most of its advantages are not valid. |
| services/backup/docker-server/contents/server/src/Reactors/server/SendLogReactor.h | ||
|---|---|---|
| 165–169 ↗ | (On Diff #11174) | I think the callback would need to be synced anyways - We cannot just call it whenever we like, at some point we have to wait until the operations on the other reactor are finished (so in the case you're describing, we'd have to wait in the other reactor, not here). |