Bug #74323 Avoid overloading the master NIC on relay-log-recovery of a lagging slave.
Submitted: 10 Oct 2014 13:38 Modified: 10 Oct 2014 13:44
Reporter: Jean-François Gagné Email Updates:
Status: Open Impact on me:
Category:MySQL Server: Replication Severity:S4 (Feature request)
Version:5.6.21 OS:Any
Assigned to: CPU Architecture:Any
Tags: Crash-safe replication, Delayed Slave, relay-log-recovery, replication

[10 Oct 2014 13:38] Jean-François Gagné
On relay-log-recovery, a slave redownloads all unexecuted relay-logs from the master.  On a mostly up to date slave, this is not a problem.  But it can be a big problem on a lagging slave: redownloading those relay-logs could overload the network interface of the master (or the WAN if the slave is on a remote site).

This is especially true for delayed slaves: a delayed slave with 10 GB of unexecuted relay-logs will overload a 1 Gb NIC for more than one minute.

Thanks in advance for improving this.

How to repeat:
See Bug #74321 for "how to repeat" overloading the master on relay-log-recovery.
[10 Oct 2014 13:44] Jean-François Gagné
Related to Bug #74089.
[10 Oct 2014 14:48] Shane Bester
Also: http://bugs.mysql.com/bug.php?id=71863 (Improve relay_log_recovery behaviour for delayed slaves)