[MDEV-22527] Add thread info for innodb semaphore wait in error log Created: 2020-05-11 Updated: 2021-09-30 Resolved: 2021-07-20 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | Storage Engine - InnoDB |
| Fix Version/s: | N/A |
| Type: | Task | Priority: | Major |
| Reporter: | Manjot Singh (Inactive) | Assignee: | Marko Mäkelä |
| Resolution: | Won't Fix | Votes: | 1 |
| Labels: | None | ||
| Issue Links: |
|
||||||||
| Description |
|
In the error log, with InnoDB semaphore wait NOTE or WARNING (as shown below), information about which thread is holding the mutex is not shown. E.g.:
The log should show information on who is holding the mutexes to aid in troubleshooting:
|
| Comments |
| Comment by Geoff Montee (Inactive) [ 2020-05-11 ] |
|
This appears to be a duplicate of |
| Comment by Geoff Montee (Inactive) [ 2020-05-11 ] |
|
This issue differs from
I agree that the thread ID of the thread holding the mutex should be printed as well. However, we still need the above information to be printed too. The thread ID is not always helpful on its own. The problem is that some threads do not appear in SHOW ENGINE INNODB STATUS, so the thread ID is not always enough to identify the thread. In that case, the information requested by |
| Comment by Geoff Montee (Inactive) [ 2020-05-21 ] |
|
My perspective of this is that 1.) Both pieces of this information appear to be necessary for proper diagnostics, because not all threads appear in the lock monitor in the error log or SHOW ENGINE INNODB STATUS output. For those threads, #2 may not be enough information to diagnose the problem. In that case, you may need #1 to determine if the semaphore is held by some internal InnoDB thread that is mostly hidden from the user. |
| Comment by Eugene Kosov (Inactive) [ 2020-08-04 ] |
|
|
| Comment by Marko Mäkelä [ 2021-07-20 ] |
|
The semaphore wait section was removed in |
| Comment by Manjot Singh (Inactive) [ 2021-07-20 ] |
|
Marko, I don't see this removal in the ticket description. Why would we remove important and critical observability from our engine? |