-
Notifications
You must be signed in to change notification settings - Fork 553
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
assertion in controller log flush #15201
Comments
From @mmaslankaprv:
|
The log entries from the node that crashed:
From the logs we can see that the write was successful as the new segment is created. The log entry: |
This is a tricky bug. We're recording a This is only possible if the data is flushed partially. I tried to reproduce this by doing concurrent log append/flush operations with embedded sleeps in various places but wasn't able to reproduce yet. |
Upd. I wasn't able to reproduce this yet. The theory is that we have a concurrent truncation and append entries request. The 'flush' picked up the first segment out of two (one just added) and flushed it expecting committed offset to move forward. But the data batch was added to the second segment (it has one batch and it's not persisted yet). This violates Raft consistency guarantees so it makes total sense to terminate. I will improve logging in case this will reproduce in CI again. |
Duplicate happened in transforms, why does this seem to only happen with wasm enabled?? |
That one was for the controller log too. |
While adding a test in #15178 one of the CI repeat runs triggered an interesting assertion on the controller log.
Full logs: https://ci-artifacts.dev.vectorized.cloud/redpanda/41890/018c17eb-df1b-4ff9-8001-afcf2c7791e4/vbuild/ducktape/results/final/DataTransformsTest/test_identity/transactional=True/124/RedpandaService-0-140696509786832/docker-rp-21/redpanda.log
Full test results: https://ci-artifacts.dev.vectorized.cloud/redpanda/41890/018c17eb-df1b-4ff9-8001-afcf2c7791e4/vbuild/ducktape/results/final/DataTransformsTest/test_identity/transactional=True/124/
The text was updated successfully, but these errors were encountered: