Bug #84224 mysql group replication bug
Submitted: 16 Dec 2016 7:16 Modified: 31 Jan 2017 5:02
Reporter: mark mark Email Updates:
Status: Can't repeat Impact on me:
None 
Category:MySQL Server: Group Replication Severity:S3 (Non-critical)
Version:5.7.17-log OS:Linux (centos 6.6)
Assigned to: MySQL Verification Team CPU Architecture:Any
Tags: group replication, memory leak

[16 Dec 2016 7:16] mark mark 
Description:
when i load 10GB data into three node mysql group replication,mysqld memory usage more and more until oom。
while my host mem 16GB，innodb buffer pool 4GB。
/var/log/messages
Dec 13 13:50:50 10-25-126-227 kernel: [ 7359.315391] mysqld invoked oom-killer: gfp_mask=0x201da, order=0, oom_adj=0, oom_score_adj=0
Dec 13 13:50:50 10-25-126-227 kernel: [ 7359.315395] mysqld cpuset=/ mems_allowed=0
Dec 13 13:50:50 10-25-126-227 kernel: [ 7359.315398] Pid: 9991, comm: mysqld Not tainted 2.6.32-431.11.25.el6.ucloud.x86_64 #1

How to repeat:
Load data as much as possible to MySQL group replication

Suggested fix:
fix it,
I guess it was a memory leak.
[31 Jan 2017 5:02] MySQL Verification Team 
Hi,

I tried loading 10G of dummy data into a 3 node group replication running on smaller boxes then yours (3x4G with 2G buffer pool) and I was not able to introduce OOM.

Any more details about your system, what exactly you did, any views, stored procedures? Can you maybe upload the full log of the crashed box.

Thanks
Bogdan
[31 Jan 2017 5:04] MySQL Verification Team 
One other thing, from your report I'm not sure what you "exactly" did? Did you load all 10G on a single node? Is that the node that died or one of the other nodes died?

Please share your full configuration from all nodes, os details, full log from all three nodes..

Thanks
Bogdan