Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.6.7
-
None
-
CentOS Stream 8 x86_64
Description
SHOW SLAVE STATUS \G
|
Master_Host: 172.31.31.10
|
Master_User: orchestrator
|
Master_Port: 3306
|
Slave_IO_Running: No
|
Slave_SQL_Running: Yes
|
Last_IO_Errno: 1593
|
Last_IO_Error: Fatal error: Failed to run 'after_read_event' hook
|
|
2022-04-11 18:00:11 83 [Note] Slave I/O thread: connected to master 'orchestrator@172.31.31.10:3306',replication starts at GTID position '1-1-338'
|
2022-04-11 18:00:11 83 [ERROR] Missing magic number for semi-sync packet, packet len: 53
|
2022-04-11 18:00:11 83 [ERROR] Slave I/O: Fatal error: Failed to run 'after_read_event' hook, Internal MariaDB error code: 1593
|
2022-04-11 18:00:11 83 [Note] Slave I/O thread exiting, read up to log 'mysql-bin-13306.000001', position 86652; GTID position 1-1-338, master 172.31.31.10:3306
|
|
I have a simple MariaDB cluster, the master(172.31.31.10), and a slave(172.31.31.12).
And I randomly drop ingress packets in master and slave by the following commands:
iptables -A INPUT -p tcp --destination-port 3306 -j DROP
|
iptables -A INPUT -p icmp -j DROP
|
also I'll randomly remove above iptables rules to simulate network jitter.
And under some circumstances, I met the Fatal error: Failed to run 'after_read_event' hook, I don't know about the root cause, it seems some kinda race condition when initializing the semi-sync replication.
Also note that, when I met this kinda error, I ran START SLAVE IO_THREAD the replication will back to normal.
Attachments
Issue Links
- relates to
-
MDEV-32551 "Read semi-sync reply magic number error" warnings on master
-
- Closed
-
Thanks for the bug report. Can you share any replication setup system variables used and perhaps the contents of the binary log at the point at which it failed.
If the binary log contents are too private to be public, then https://mariadb.com/kb/en/meta/mariadb-ftp-server/ is available for communication to developers directly.