Relay log write failure

Hello

Today I've got a very strange behaviour on replicas that works under maxscale binlogloter

I have next schema:
2 masters (main and reserve) --> Maxscale binlog --> Few replicas

I did planned task of switch master from main to reserve
And all replicas under maxscale were stopped with error

Last_IO_Error: Relay log write failure: could not queue event from master

Mysql log return next error

2024-10-11  7:10:55 207183263 [ERROR] Unexpected break of being relay-logged GTID 11-22-148858610346 event group by the current GTID event 11-22-148858610347

2024-10-11  7:10:55 207183263 [ERROR] Slave I/O: Relay log write failure: could not queue event from master, Internal MariaDB error code: 1595

Look like mariadb replica can't get event from binlogrouter and write it to own relay log.

At first for solve the problem I tried restart replication (stop slave; start slave; ). This not helped.
Replication has been restored only when I re-configured replication ( on each replica ) directly to reserve master. And only after that action I could return replicas back to maxscale binlogrouter.

Chronology at log from replica (mariadb.log)
7:10:55 switch master
7:19:21 restart replica
7:20:58 connect to reserve master
7:27:40 return to maxscale binlog

I've checked binlog files on maxscale binlog router . They are looks correct , only one issue that 22-148858610345 was a last transaction in file mysqld-bin.443471
I've attached cropped logs (few events in each binlog file and I masked personal data by *****)
Also I attached log from one of replicas (other replicas have the same logs) and from maxscale binlogrouter