Bug #89115 | Unnecessary to hold trx mutex while updating trx_t::n_ref | ||
---|---|---|---|
Submitted: | 5 Jan 2018 10:05 | Modified: | 15 Jan 2018 3:23 |
Reporter: | zhai weixiang (OCA) | Email Updates: | |
Status: | Verified | Impact on me: | |
Category: | MySQL Server: InnoDB storage engine | Severity: | S4 (Feature request) |
Version: | MySQL8.0 | OS: | Any |
Assigned to: | CPU Architecture: | Any |
[5 Jan 2018 10:05]
zhai weixiang
[12 Jan 2018 13:32]
MySQL Verification Team
Hi! Thank you for your bug report. Your analysis is oversimplified, but, it is also relatively correct. The only bad consequence of this design is that several of the mutex locks are overused, which increases contention and that is not what we want. Hence, this makes a good feature request and I am verifying it as such.
[12 Jan 2018 17:30]
MySQL Verification Team
Hi! It turns out that trx_sys contention is known, but we have not seen a contention on the trx_t mutex. Can you underline scenario for that contention, please .......
[15 Jan 2018 3:23]
zhai weixiang
Hi, Sinisa Thank you for verifying this bug. I didn't see hot mutex contnetion on trx_t::mutex while running benchmark. I found this issue while investigating bug #89127