apache performance autotuning fails on multi core system

Bug #1365584 reported by Julian Klinck
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Fuel for OpenStack
Confirmed
High
Aleksandr Didenko

Bug Description

We found out that a fresh deployment of a FUEL controller node with CEPH onto a 24-Core machine with 32GB of memory resulted in a none working object storage. The issue was that the rados-gw service didn't came up "Operation not permitted".

The issue is that apache is too aggressively auto tuned by the script:
https://github.com/stackforge/fuel-library/blob/master/deployment/puppet/horizon/templates/zzz_performance_tuning.conf.erb

The number of servers and max_clients are such high that we run out of different hard centos system limits (open files, max threads). Reducing the amount of start_servers made rados-gw service working.

Either introduce hard upper boundaries for such auto tuning or increase the system limits, too.

Tags: ceph
description: updated
description: updated
Dmitry Pyzhov (dpyzhov)
Changed in fuel:
milestone: none → 5.1
assignee: nobody → Fuel Library Team (fuel-library)
Changed in fuel:
assignee: Fuel Library Team (fuel-library) → Aleksandr Didenko (adidenko)
Changed in fuel:
importance: Undecided → High
status: New → Confirmed
Revision history for this message
Dmitry Borodaenko (angdraug) wrote :

Looks like a duplicate of bug #1331554.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.