Skip to content

Conversation

@enjoy-binbin
Copy link
Contributor

The new blockedBeforeSleep was added in #12337, it breaks the order in 2ecb5ed.

This may be related to #2288, quoted from comment in #2288:

Moreover the clusterBeforeSleep() call was misplaced at the end of the chain of the
beforeSleep() call in redis.c. It should be at the top, before processing un blocking
clients. This is exactly the reason in the specific instance of the bug as reported,
why the state was not updated in time before clients served.

The new blockedBeforeSleep was added in redis#12337, it breaks the order in 2ecb5ed.

This may be related to redis#2288, quoted from comment in redis#2288:
```
Moreover the clusterBeforeSleep() call was misplaced at the end of the chain of the
beforeSleep() call in redis.c. It should be at the top, before processing un blocking
clients. This is exactly the reason in the specific instance of the bug as reported,
why the state was not updated in time before clients served.
```
@enjoy-binbin
Copy link
Contributor Author

another question: should cron_start_time_before_aof be considered?

@oranagra
Copy link
Member

another question: should cron_start_time_before_aof be considered?

i don't think clusterBeforeSleep is expected to consume a lot of time, i don't mind if we don't count it.

@enjoy-binbin
Copy link
Contributor Author

i can't reproduce the problem and write a test case now, please feel free to merge it first.
i will find another day to try

Co-authored-by: Binbin <binloveplay1314@qq.com>
@oranagra oranagra merged commit 9600553 into redis:unstable Jun 25, 2023
@enjoy-binbin enjoy-binbin deleted the block_order branch June 25, 2023 11:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: Done

Development

Successfully merging this pull request may close these issues.

2 participants