After update mariadb server to version 10.4.20, replication stops on random(usualy different) slave with error code: 1032 and different table every few hours but every other slave run without problem. Same with mariadb 10.4.22 (10.4.21 not tested), downgraded master back to 10.4.19 resolve the issue, slave has no problem with 10.4.20,21,22.

I increased log verbosity to log_warnings=3 but still only following is logged to error_log on slaves in that time window:

Slave SQL: Could not execute Delete_rows_v1 event on table d_xfovvr2vr59rjd.jos_jf_content; Can't find record in 'jos_jf_content', Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND; the event's master log binary_log.010420, end_log_pos 129891745, Gtid 1-13-2267955079, Internal MariaDB error code: 1032

Slave SQL: Could not execute Update_rows_v1 event on table d_7y5z7ku1s63yp5.jos_vm_coupons; Can't find record in 'jos_vm_coupons', Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND; the event's master log binary_log.010421, end_log_pos 233244210, Gtid 1-13-2268995146, Internal MariaDB error code: 1032

Nothing unusual on masters.

There was no problem with version 10.4.12 to 10.4.19 on master.

Setup contains 3 master servers (only one so active at time) and 6 slaves servers(all active), all slaves and also inactive masters are read only. Master had read only unsetted onlye after all events all aplayed and no regular user has SUPER privilege.

Master and slave cnf added to attachment, removed all comments.