Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.0.24
-
None
-
Debian Linux
Description
We have a MariaDB server that sometimes seems to get an indefinite table lock. The only way to solve it is to kill MariaDB and restart it. It "randomly happens", so I suspect it may be a race condition. The replication thread get's stuck in state "After opening tables". The lock seems to be caused by a query on an unrelated database with 3 million MyISAM tables (not receiving data via replication) in state "Opening tables". These databases are completely separated, so they shouldn't cause a lock between each other.
Processlist shows:
root@server ~ # mysql -e "show full processlist" |
+----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ |
| Id | User | Host | db | Command | Time | State | Info | Progress | |
+----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ |
| 2 | system user | | NULL | Connect | 1116538 | Waiting for master to send event | NULL | 0.000 | |
| 3 | system user | | replication_database | Connect | 264121 | After opening tables | COMMIT | 0.000 | |
| 37332419 | user1 | localhost | large_database | Killed | 264121 | Opening tables | DELETE FROM `my_domain` WHERE date<'2016-04-29' | 0.000 | |
| 48692447 | root | localhost | NULL | Query | 0 | init | show full processlist | 0.000 | |
+----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ |
root@server ~ #
|
Attachments
Activity
| Field | Original Value | New Value |
|---|---|---|
| Description |
We have a MariaDB server that sometimes seems to get an indefinite table lock. The only way to solve it is to kill MariaDB and restart it. It "randomly happens", so I suspect it may be a race condition. The replication thread get's stuck in state "After opening tables". The lock seems to be caused by a query on an unrelated database with 3 million MyISAM tables (not receiving data via replication) in state "Opening tables". These databases are completely separated, so they shouldn't cause a lock between each other.
Processlist shows: === root@server ~ # mysql -e "show full processlist" +----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ | Id | User | Host | db | Command | Time | State | Info | Progress | +----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ | 2 | system user | | NULL | Connect | 1116538 | Waiting for master to send event | NULL | 0.000 | | 3 | system user | | replication_database | Connect | 264121 | After opening tables | COMMIT | 0.000 | | 37332419 | user1 | localhost | large_database | Killed | 264121 | Opening tables | DELETE FROM `my_domain` WHERE date<'2016-04-29' | 0.000 | | 48692447 | root | localhost | NULL | Query | 0 | init | show full processlist | 0.000 | +----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ root@server ~ # === |
We have a MariaDB server that sometimes seems to get an indefinite table lock. The only way to solve it is to kill MariaDB and restart it. It "randomly happens", so I suspect it may be a race condition. The replication thread get's stuck in state "After opening tables". The lock seems to be caused by a query on an unrelated database with 3 million MyISAM tables (not receiving data via replication) in state "Opening tables". These databases are completely separated, so they shouldn't cause a lock between each other.
Processlist shows: {code:sql} root@server ~ # mysql -e "show full processlist" +----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ | Id | User | Host | db | Command | Time | State | Info | Progress | +----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ | 2 | system user | | NULL | Connect | 1116538 | Waiting for master to send event | NULL | 0.000 | | 3 | system user | | replication_database | Connect | 264121 | After opening tables | COMMIT | 0.000 | | 37332419 | user1 | localhost | large_database | Killed | 264121 | Opening tables | DELETE FROM `my_domain` WHERE date<'2016-04-29' | 0.000 | | 48692447 | root | localhost | NULL | Query | 0 | init | show full processlist | 0.000 | +----------+----------------------+-----------------------+----------------------+---------+---------+----------------------------------+-------------------------------------------------------+----------+ root@server ~ # {code} |
| Labels | need_feedback |
| Labels | need_feedback |
| Labels | need_feedback |
| Fix Version/s | N/A [ 14700 ] | |
| Resolution | Incomplete [ 4 ] | |
| Status | Open [ 1 ] | Closed [ 6 ] |
| Attachment | global_status.txt [ 42928 ] |
| Assignee | Elena Stepanova [ elenst ] | |
| Resolution | Incomplete [ 4 ] | |
| Status | Closed [ 6 ] | Stalled [ 10000 ] |
| Assignee | Elena Stepanova [ elenst ] |
| Fix Version/s | N/A [ 14700 ] |
| Labels | need_feedback |
| Assignee | Elena Stepanova [ elenst ] |
| Attachment | strace.txt [ 43385 ] |
| Attachment | strace_170419_2.txt [ 43561 ] | |
| Attachment | strace_170419_1.txt [ 43562 ] |
| Labels | need_feedback |
| Assignee | Elena Stepanova [ elenst ] | Michael Widenius [ monty ] |
| Attachment | strace_170425.txt [ 43581 ] |
| Labels | need_feedback |
| Attachment | patch [ 43617 ] |
| Status | Stalled [ 10000 ] | In Progress [ 3 ] |
| issue.field.resolutiondate | 2017-05-05 15:27:57.0 | 2017-05-05 15:27:57.607 |
| Component/s | Storage Engine - Aria [ 10126 ] | |
| Component/s | Locking [ 10900 ] | |
| Fix Version/s | 10.0.31 [ 22501 ] | |
| Resolution | Fixed [ 1 ] | |
| Status | In Progress [ 3 ] | Closed [ 6 ] |
| Workflow | MariaDB v3 [ 75740 ] | MariaDB v4 [ 150447 ] |