Bug #87652 InnoDB: Warning: a long semaphore wait:
Submitted: 4 Sep 2017 9:35 Modified: 4 Sep 2017 19:40
Reporter: Dominik Müller Email Updates:
Status: Can't repeat Impact on me:
None 
Category:MySQL Server: InnoDB storage engine Severity:S1 (Critical)
Version:mysql Ver 14.14 Distrib 5.5.57, for deb OS:Debian (Jessie)
Assigned to: Bogdan Kecman CPU Architecture:Any
Tags: innodb semaphore crashs

[4 Sep 2017 9:35] Dominik Müller
Description:
This Error happens every few days. It is a shared hosting Server with only about 15 customers on it. The Databases that lead to this error seems to be owncloud/nextcloud-Installations.

The error leads to a crash of mysql, and a long restart time when it is manually (or automatically by monitoring) restarted. In meanwhile the mysql-service is not usable, not reachable.

How to repeat:
not known, it just happens on this one system (I have running several systems with same configuration and even much more customers)
[4 Sep 2017 19:40] Bogdan Kecman
Hi,

There is no crash in the log you provided, only a shutdown/restart and bunch of errors accessing disk during startup.. probably your startup scripts are not ok but since this does not look like mysql binary those scripts are also not provided by us but probably by your distro.

InnoDB monitor log does show congestion but this does not look like a bug but a problem with storage subsystem (maybe that shared hosting has slow disks?, low iops capacity?), in any way this is something you can sort out with our Support team and properly configure your system, I don't see a bug here.

best regards
Bogdan