Details
-
Type:
Bug
-
Status: Open
-
Priority:
Minor
-
Resolution: Unresolved
-
Affects Version/s: 10.0.15
-
Fix Version/s: 10.0
-
Component/s: Replication
-
Labels:
Description
Ever since migrating to mariadb with named masters; besides useful relay logs (named mysql-relay-bin-SERVER which are properly removed) we have a lot of useless, empty relay logs on all our servers. It seems every FLUSH LOGS (which we execute daily) results in an additional file. They are all 297 bytes.
These shouldn't be created since we only have named masters.
Gliffy Diagrams
Attachments
Issue Links
- relates to
-
MDEV-8596 Detect conflicts between file name settings of similarly named variables
-
- Open
-
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
I already checked that; and I really only have 2 named masters there. This machine did have an unnamed connection previously; but that is multiple months, reboots and upgrades ago.
Could you try if "flush logs" maybe causes these files to appear? I'm pretty sure it's related to that:
-rw-rw---- 1 mysql mysql 13470762 Jan 1 00:20 mysql-bin.000081 -rw-rw---- 1 mysql mysql 13573801 Jan 2 00:34 mysql-bin.000082 -rw-rw---- 1 mysql mysql 12280631 Jan 3 00:41 mysql-bin.000083 -rw-rw---- 1 mysql mysql 7220622 Jan 4 00:41 mysql-bin.000084 -rw-rw---- 1 mysql mysql 4100754 Jan 5 00:10 mysql-bin.000085 -rw-rw---- 1 mysql mysql 20281672 Jan 6 00:25 mysql-bin.000086 -rw-rw---- 1 mysql mysql 19206399 Jan 6 15:12 mysql-bin.000087 -rw-rw---- 1 mysql mysql 603780 Jan 6 17:35 mysql-bin.000088 -rw-rw---- 1 mysql mysql 345 Jan 6 17:57 mysql-bin.000089 -rw-rw---- 1 mysql mysql 183694 Jan 7 00:17 mysql-bin.000090 -rw-rw---- 1 mysql mysql 414241 Jan 7 09:30 mysql-bin.000091 -rw-rw---- 1 mysql mysql 70131 Jan 7 09:42 mysql-bin.000092 -rw-rw---- 1 mysql mysql 5086 Jan 7 09:56 mysql-bin.000093 -rw-rw---- 1 mysql mysql 345 Jan 7 10:10 mysql-bin.000094 -rw-rw---- 1 mysql mysql 369 Jan 8 00:20 mysql-bin.000095 -rw-rw---- 1 mysql mysql 2816 Jan 8 15:40 mysql-bin.000096 -rw-rw---- 1 mysql mysql 512 Jan 8 00:20 mysql-bin.index ...skipped 179 files, removed named relay-logs ... -rw-rw---- 1 mysql mysql 297 Jan 1 00:20 mysql-relay-bin.000180 -rw-rw---- 1 mysql mysql 297 Jan 2 00:34 mysql-relay-bin.000181 -rw-rw---- 1 mysql mysql 297 Jan 3 00:41 mysql-relay-bin.000182 -rw-rw---- 1 mysql mysql 297 Jan 4 00:41 mysql-relay-bin.000183 -rw-rw---- 1 mysql mysql 297 Jan 5 00:10 mysql-relay-bin.000184 -rw-rw---- 1 mysql mysql 297 Jan 6 00:25 mysql-relay-bin.000185 -rw-rw---- 1 mysql mysql 267 Jan 6 15:12 mysql-relay-bin.000186 -rw-rw---- 1 mysql mysql 267 Jan 6 17:35 mysql-relay-bin.000187 -rw-rw---- 1 mysql mysql 267 Jan 6 17:57 mysql-relay-bin.000188 -rw-rw---- 1 mysql mysql 297 Jan 7 00:17 mysql-relay-bin.000189 -rw-rw---- 1 mysql mysql 267 Jan 7 09:30 mysql-relay-bin.000190 -rw-rw---- 1 mysql mysql 267 Jan 7 09:42 mysql-relay-bin.000191 -rw-rw---- 1 mysql mysql 267 Jan 7 09:56 mysql-relay-bin.000192 -rw-rw---- 1 mysql mysql 267 Jan 7 10:10 mysql-relay-bin.000193 -rw-rw---- 1 mysql mysql 297 Jan 8 00:20 mysql-relay-bin.000194 -rw-rw---- 1 mysql mysql 248 Jan 8 00:20 mysql-relay-bin.000195 -rw-rw---- 1 mysql mysql 4875 Jan 8 00:20 mysql-relay-bin.index
Yes, I did run FLUSH LOGS, the anonymous logs only appear for me when the anonymous slave is configured.
Do you have other files related to the anonymous slave in your datadir? master.info, relay-log.info.
There should also be the file multi-master.info, can you check its contents – does it only have two lines corresponding to your configured slaves, or is there anything else?
Please also attach output of SHOW VARIABLES and SHOW GLOBAL VARIABLES.
Thanks.
multi-master contains 2 lines which is correct; interestingly I also have the master.info which seems to be somewhat odd:
33
4
3306
60
1
/etc/mysql/ca-cert.pem0
1800.0000
0
using_gtid=0
The relay-log.info is also there:
./mysql-relay-bin.000001
40
I'll attach the output of the variables
Thanks, I could now reproduce the problem.
It starts happening if you at least once attempted to run START SLAVE (without a name), while the anonymous slave is not configured. The attempt fails with the proper error, as expected, and the slave does not appear in SHOW ALL SLAVES STATUS, but all files associated with it (relay logs etc.) get created and from now on start rotating etc.
Test case
--source include/master-slave.inc --connection slave --source include/stop_slave.inc reset slave all; eval change master 'm1' to master_host='127.0.0.1', master_port=$MASTER_MYPORT, master_user='root'; --let $datadir = `select @@datadir` start all slaves; set default_master_connection = 'm1'; --source include/wait_for_slave_to_start.inc set default_master_connection = ''; --echo # --echo # Relay logs after starting 'm1' slave --list_files $datadir *relay* --echo # --echo # FLUSH LOGS doesn't do anything bad flush logs; --list_files $datadir *relay* --echo # --echo # Try to start the anonymous slave although it's not configured --error ER_BAD_SLAVE start slave; --echo # --echo # Relay logs after a failed attempt to start an unconfigured slave --list_files $datadir *relay* --echo # --echo # From now on FLUSH LOGS rotate logs for non-existing slave flush logs; --list_files $datadir *relay*
Partial output
# # Relay logs after starting 'm1' slave relay-log-m1.info slave-relay-bin-m1.000001 slave-relay-bin-m1.000002 slave-relay-bin-m1.index # # FLUSH LOGS doesn't do anything bad flush logs; relay-log-m1.info slave-relay-bin-m1.000001 slave-relay-bin-m1.000002 slave-relay-bin-m1.index # # Try to start the anonymous slave although it's not configured start slave; ERROR HY000: The server is not configured as slave; fix in config file or with CHANGE MASTER TO # # Relay logs after a failed attempt to start an unconfigured slave relay-log-m1.info relay-log.info slave-relay-bin-m1.000001 slave-relay-bin-m1.000002 slave-relay-bin-m1.index slave-relay-bin.000001 slave-relay-bin.index # # From now on FLUSH LOGS rotate logs for non-existing slave flush logs; relay-log-m1.info relay-log.info slave-relay-bin-m1.000001 slave-relay-bin-m1.000002 slave-relay-bin-m1.index slave-relay-bin.000001 slave-relay-bin.000002 slave-relay-bin.index
Strangely, I don't get these default relay logs if I only ever configure named slaves.
Could it be that you once ran change master to ..., even though you never actually started this anonymous slave? It would behave exactly as you described.
The easy way to check it is to run SHOW ALL SLAVES STATUS \G command. It will return a row per each configured slave; the first field is 'Connection_name' – if it's empty, it is the anonymous slave.
If it's there and you don't want it, you can run RESET SLAVE '' ALL.