SLEEP thread causing "Waiting for table metadata lock"

By

SLEEP thread causing "Waiting for table metadata lock"

SLEEP thread causing "Waiting for table metadata lock"

We might have faced some scenarios once our db connection threads are getting locked and we try to find the culprit thread id which is locking the thread, to kill and release the connections. 

Here I am going to focus on one of the scenarios when db connection threads are locked at state "waiting for table metadata lock" and when we see the processlist, we can't find any thread which is running and using the same table, i.e., all threads are in SLEEP state except the locked threads.


Such as : 

mysql [localhost:5726] {msandbox} (world) > show processlist\G
*************************** 1. row ***************************
           Id: 5
         User: rsandbox
         Host: localhost:60624
           db: NULL
      Command: Binlog Dump
         Time: 111519
        State: Master has sent all binlog to slave; waiting for more updates
         Info: NULL
    Rows_sent: 0
Rows_examined: 0
*************************** 2. row ***************************
           Id: 9
         User: msandbox
         Host: localhost
           db: NULL
      Command: Sleep
         Time: 1901
        State:
         Info: NULL
    Rows_sent: 1
Rows_examined: 0
*************************** 3. row ***************************
           Id: 10
         User: msandbox
         Host: localhost
           db: world
      Command: Query
         Time: 282
        State: Waiting for table metadata lock
         Info: alter table city engine=innodb
    Rows_sent: 0
Rows_examined: 0
*************************** 4. row ***************************
           Id: 13
         User: msandbox
         Host: localhost
           db: NULL
      Command: Sleep
         Time: 645
        State:
         Info: NULL
    Rows_sent: 0
Rows_examined: 0
*************************** 5. row ***************************
           Id: 14
         User: msandbox
         Host: localhost
           db: world
      Command: Query
         Time: 0
        State: starting
         Info: show processlist
    Rows_sent: 0
Rows_examined: 0
*************************** 6. row ***************************
           Id: 17
         User: msandbox
         Host: localhost
           db: world
      Command: Sleep
         Time: 285
        State:
         Info: NULL
    Rows_sent: 0
Rows_examined: 4079
6 rows in set (0.00 sec)

mysql [localhost:5726] {msandbox} (world) >
                                                       Below I will explain about how we can troubleshoot this issue, release the locked thread and also get the transaction details of thread which was present in SLEEP state and was causing the "waiting for table metadata lock" to other threads.


Above we can see our thread id 10 is locked at state "waiting for table metadata lock" but on processlist, we can't see any other active thread which is running for table city. All other threads are present in the SLEEP state.

Here We can use below ways to find the culprit thread id which is locking the thread id 10.

A) Using InnoDB Engine Status 

-- Once we will run "show engine innodb status", here we can see there is one active transaction/thread present, and it has "24 lock struct(s), heap size 3520, 4102 row lock(s), undo log entries 1". 

mysql [localhost:5726] {msandbox} (world) > show engine innodb status\G
*
*
*
*
------------
TRANSACTIONS
------------
Trx id counter 1876
Purge done for trx's n:o < 1875 undo n:o < 0 state: running but idle
History list length 24
LIST OF TRANSACTIONS FOR EACH SESSION:
---TRANSACTION 421869282924072, not started
0 lock struct(s), heap size 1136, 0 row lock(s)
---TRANSACTION 421869282922944, not started
0 lock struct(s), heap size 1136, 0 row lock(s)
---TRANSACTION 421869282921816, not started
0 lock struct(s), heap size 1136, 0 row lock(s)
---TRANSACTION 1875, ACTIVE 137 sec
24 lock struct(s), heap size 3520, 4102 row lock(s), undo log entries 1
MySQL thread id 17, OS thread handle 140394578265856, query id 5478 localhost msandbox
Trx read view will not see trx with id >= 1875, sees < 1875
*
*
*
mysql [localhost:5726] {msandbox} (world) > 

B) Using the information_schema database tables

-- We can also find the culprit thread id using the INFORMATION_SCHEMA.INNODB_TRX table.

mysql [localhost:5726] {msandbox} (information_schema) > select ps.id,ps.user,ps.host,ps.db,ps.command from information_schema.processlist ps join information_schema.INNODB_TRX itx on itx.trx_mysql_thread_id=ps.id and ps.command='Sleep';
+----+----------+-----------+-------+---------+
| id | user     | host      | db    | command |
+----+----------+-----------+-------+---------+
| 17 | msandbox | localhost | world | Sleep   |
+----+----------+-----------+-------+---------+
1 row in set (0.00 sec) 

By running the above query, we can see we have thread id 17 which is active under the INFORMATION_SCHEMA.INNODB_TRX table and currently present in SLEEP state on processlist.

C) Using information_schema and performance_schema database tables

Here it is assumed we have performance_schema enabled. 

-- By running the below query we can check the current active/open transaction as well as the list of queries executed by the current active/open transaction which will help the developer to troubleshoot the application effectively once they have the culprit thread detail with the list of queries executed by the open transaction. 

mysql [localhost:5726] {msandbox} ((none)) > SELECT   pslst.id 'PROCESS ID'  ,itrx.trx_id 'TRX_ID'     ,pesh.event_name 'EVENT NAME'     ,pesh.sql_text 'SQL'     ,th.processlist_user 'USER'     ,th.processlist_host 'HOST' FROM   information_schema.innodb_trx itrx JOIN information_schema.processlist pslst ON   itrx.trx_mysql_thread_id = pslst.id JOIN performance_schema.threads th ON   th.processlist_id = itrx.trx_mysql_thread_id JOIN performance_schema.events_statements_history pesh ON   pesh.thread_id = th.thread_id WHERE   itrx.trx_started < CURRENT_TIME - INTERVAL 100 SECOND  AND pslst.USER != 'SYSTEM_USER' GROUP BY  pesh.sql_text ORDER BY   pesh.EVENT_ID\G
*************************** 1. row ***************************
PROCESS ID: 17
    TRX_ID: 1875
EVENT NAME: statement/sql/select
       SQL: SELECT DATABASE()
      USER: msandbox
      HOST: localhost
*************************** 2. row ***************************
PROCESS ID: 17
    TRX_ID: 1875
EVENT NAME: statement/com/Init DB
       SQL: NULL
      USER: msandbox
      HOST: localhost
*************************** 3. row ***************************
PROCESS ID: 17
    TRX_ID: 1875
EVENT NAME: statement/sql/show_databases
       SQL: show databases
      USER: msandbox
      HOST: localhost
*************************** 4. row ***************************
PROCESS ID: 17
    TRX_ID: 1875
EVENT NAME: statement/sql/show_tables
       SQL: show tables
      USER: msandbox
      HOST: localhost
*************************** 5. row ***************************
PROCESS ID: 17
    TRX_ID: 1875
EVENT NAME: statement/sql/begin
       SQL: start transaction
      USER: msandbox
      HOST: localhost
*************************** 6. row ***************************
PROCESS ID: 17
    TRX_ID: 1875
EVENT NAME: statement/sql/select
       SQL: select * from city limit 2
      USER: msandbox
      HOST: localhost
*************************** 7. row ***************************
PROCESS ID: 17
    TRX_ID: 1875
EVENT NAME: statement/sql/update
       SQL: update city set Population=1780001 where Name='Kabul'
      USER: msandbox
      HOST: localhost
7 rows in set (0.01 sec)


mysql [localhost:5726] {msandbox} ((none)) > 






Above we can see we have active/open transaction id 17 which have started the transaction but not yet committed, and currently, we can see this thread is present in SLEEP state on processlist.


                                       So above we saw a couple of ways where we can find and confirm the culprit thread id which is causing for the "waiting for table metadata lock" issue.


Now once we have culprit thread id details from above, We can now KILL the culprit thread id and we can release the locked thread id.
mysql [localhost:5726] {msandbox} ((none)) > kill 17;
Query OK, 0 rows affected (0.01 sec)
mysql [localhost:5726] {msandbox} (world) > alter table city engine=innodb;
Query OK, 0 rows affected (26 min 8.17 sec)
Records: 0  Duplicates: 0  Warnings: 0
Hope this blog post will help to troubleshoot the "waiting for table metadata lock" issue once we have all other threads present in SLEEP state on processlist. 


 Photo by drmakete lab on Unsplash

Comments

Post a Comment

Popular

MySQL Memory Calculator

By
Image
MySQL Memory Calculator Check out the “ MySQL Memory Calculator ” added to my blog page, which will be helpful to calculate MySQL memory usage during tuning of MySQL memory parameters. https://www.abhinavbit.com/p/mysql-memory-calculator.html Photo by  Kelly Sikkema  on  Unsplash
Read more

errno: 24 - Too many open files

By
Image
errno: 24 - Too many open files You might see this error inside MySQL's error log file or during mysql_upgrade operation. Can't open file: ‘./db/table.frm' (errno: 24 - Too many open files) To solve this issue, you will need to change the open files limit at system and db service level. --> To check and change this limit at the system level, we use ulimit client tool to set it dynamically. * Check current system open file limit. root@test:~# ulimit -a core file size          (blocks, -c) 0 data seg size           (kbytes, -d) unlimited scheduling priority             (-e) 0 file size               (blocks, -f) unlimited pending signals                 (-i) 515256 max locked memory       (kbytes, -l) 64 max memory size         (kbytes, -m) unlimited open files                      (-n) 1024 pipe size            (512 bytes, -p) 8 POSIX message queues     (bytes, -q) 819200 real-time priority              (-r) 0 stack size     
Read more

How to properly shutdown MySQL before any maintenance activity

By
Image
How to properly shutdown MySQL before any maintenance activity We might have different scenarios once we need to stop MySQL service before performing either server or database activity like patching/upgrade.   Before performing such activity, we need to make sure that we stop MySQL service properly to avoid any unforeseen crashes once our maintenance activity complete and MySQL service is started.  Here I would like to share the below steps which should be performed and take care to properly stop MySQL service.  Step 1: Ensure we don't have any long-running queries. Manually verify by checking the processlist. show processlist;  Step 2: If it is the master node, then ensure to stop application write hits on the database by doing either stop an app and put in maintenance OR if we have any failover solution then move write hits on another node.  Step 3: If it is a slave node, then STOP replication and note down the stopped replication co
Read more

How to create CentOS VM using Vagrant

By
Image
Create CentOS VM using Vagrant 1. Download and install the latest Oracle VM VirtualBox.  https://www.oracle.com/technetwork/server-storage/virtualbox/downloads/index.html Like (For Mac OS X) https://download.oracle.com/virtualbox/6.0.6/VirtualBox-6.0.6-130049-OSX.dmg 2. Download and install the latest Vagrant. https://www.vagrantup.com/downloads.html Like (For Mac OS X) https://releases.hashicorp.com/vagrant/2.2.4/vagrant_2.2.4_x86_64.dmg 3. Download " centos /7" Vagrant Box using below command. abhinavs -Air : ~ agupta$ vagrant box add centos /7 ==> box: Loading metadata for box ' centos /7'      box : URL: https://vagrantcloud.com/centos/7 This box can work with multiple providers! The providers that it can work with are listed below. Please review the list and choose the provider you will be working with. 1) hyperv 2) libvirt 3) virtualbox 4) vmware_desktop Enter your choice: 3 ==> box: Addin
Read more

ERROR 1040 (HY000): Too many connections

By
Image
ERROR 1040 (HY000): Too many connections An application might face this issue when all the defined max_connections parameter value is exhausted.  mysqld reserve a  +1 connection (i.e., max_connections + 1) for the user who is having the SUPER privilege.  When this issue appears, Only the user with the SUPER privileges can be connected if any other user who has SUPER privileges is not already connected to the database and we have reached max_connections.  1. Always make sure to DON'T give the SUPER privilege to any application user or any user which is being used inside the scripts. Because if that user has already been connected to db when "too many connections" issue appears, then that time root user can't be connected even though root user is having all privileges along with SUPER privileges.  Here I am presenting 3 cases which will help us, how MySQL handles "too many connections" and reserve 1 connection. For below example,
Read more

set --server-id to enable either a master or a slave

By
set --server-id to enable either a master or a slave Below issue can be faced, when we run CHANGE MASTER command to setup replication. mysql> change master to master_host='xx.xx.xx.xx',master_user='rep_user',master_password="xxxxxxx", master_log_file='db-01-test-bin.000064',master_log_pos=95854072,master_auto_position=0; ERROR 1794 (HY000): Slave is not configured or failed to initialize properly. You must at least set --server-id to enable either a master or a slave. Additional error messages can be found in the MySQL error log. Solution :- Here It is assumed, We have already set server-id parameter value different on master and slave node but still, we have the above error. It happens once, you performed the server packages upgrade(like from 5.6 to 5.7) and just started MySQL service and going to run CHANGE MASTER command. So, to solve this issue, Run mysql_upgrade command to update/upgrade system tables and then restart M
Read more

GTID Replication (Skip Transaction using empty transaction)

By
Image
GTID Replication (Skip Transaction using empty transaction) To skip SQL thread's error in GTID replication setup, Insert empty transaction. PROBLEM :  db -test (none ) > show slave status\G *************************** 1. row ***************************                Slave_IO_State: Waiting for master to send event                   Master_Host: xxxxx                   Master_User: rep_user                   Master_Port: 3306                 Connect_Retry: 60               Master_Log_File: db -test-02-bin . 006725           Read_Master_Log_Pos: 4508785                Relay_Log_File: db -test-relay-bin . 000015                 Relay_Log_Pos: 2146333         Relay_Master_Log_File: db -test-02-bin . 006725              Slave_IO_Running: Yes             Slave_SQL_Running: No               Replicate_Do_DB:           Replicate_Ignore_DB:            Replicate_Do_Table:        Replicate_Ignore_Table:       Replicate_Wild_Do_Table:   Replicate_Wild_Igno
Read more

Master has purged binary logs containing GTIDs that the slave requires

By
Master has purged binary logs containing GTIDs that the slave requires Problem :  Last_IO_Error: Got fatal error 1236 from master when reading data from binary log: 'The slave is connecting using CHANGE MASTER TO MASTER_AUTO_POSITION = 1, but the master has purged binary logs containing GTIDs that the slave requires.' Solution : stop slave; Query OK, 0 rows affected (0.00 sec) CHANGE MASTER TO MASTER_AUTO_POSITION = 0; Query OK, 0 rows affected (0.01 sec) start slave; Query OK, 0 rows affected (0.00 sec)
Read more

How to Prepare for MySQL Certification Exam

By
Image
How to Prepare for MySQL Certification Exam Here I am going to suggest MySQL DBA, student or any newbie that how they can prepare for "Oracle's MySQL 5.x Database Administrator" certification exam.  Step 1: Lab -- Prepare your local MySQL lab environment and make sure to install the respective MySQL version on your lab. (i.e., if you are going to appear on MySQL 5.6 DBA exam 1Z0-883 then install MySQL 5.6 on your lab)  -- You can follow my below blog post to create your local MySQL lab environment.  How to create CentOS VM using Vagrant Install MySQL on CentOS using DBdeployer Step 2: Book + HandsOn Experience -- If you don't have below book, then go right now and purchase this book. (Here Please, you don't judge that below-suggested book is for 5.0 version whereas you are going to prepare for MySQL 5.6/5.7 DBA certification exam. Just go ahead and purchase below book. In the end, you will definitely say it was your
Read more