Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.5.9
-
None
-
None
-
None
-
CentOS Linux release 8.3.2011
4.18.0-240.15.1.el8_3.x86_64
The data is stored on NVMe disk.
Description
We are facing strange performance issues after upgrading our DBMS to 10.5.9
Sometimes, with no obvious reasons, simple update queries get very slow, over 10 seconds.
The slow query log shows entries like the following:
# User@Host: user[user] @ [ip]
|
# Thread_id: 1000143 Schema: test QC_hit: No
|
# Query_time: 13.510165 Lock_time: 0.000017 Rows_sent: 0 Rows_examined: 2
|
# Rows_affected: 1 Bytes_sent: 59
|
SET timestamp=1615541184;
|
update users set last_act = now(), some_field = 0 where primary_key = 99999;
|
...
|
# Time: 700101 3:00:31
|
# User@Host: user[user] @ [ip]
|
# Thread_id: 1004560 Schema: test QC_hit: No
|
# Query_time: 31.031085 Lock_time: 0.000000 Rows_sent: 0 Rows_examined: 0
|
# Rows_affected: 0 Bytes_sent: 18446744073709547359
|
SET timestamp=31;
|
# administrator command: Change user;
|
So far I can't quite understand how to debug this situation. The server runs ~90 updates per second.