Bug 1295835

Summary:	Increase timeout defaults for pacemaker
Product:	Red Hat OpenStack	Reporter:	Michele Baldessari <michele>
Component:	openstack-tripleo-heat-templates	Assignee:	Marios Andreou <mandreou>
Status:	CLOSED ERRATA	QA Contact:	Ofer Blaut <oblaut>
Severity:	high	Docs Contact:
Priority:	urgent
Version:	7.0 (Kilo)	CC:	dmacpher, ggillies, hbrock, jason.dobies, jraju, jslagle, mburns, michele, rhel-osp-director-maint, rscarazz, sasha, sclewis, yeylon
Target Milestone:	y3
Target Release:	7.0 (Kilo)
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:	openstack-tripleo-heat-templates-0.8.6-98.el7ost	Doc Type:	Bug Fix
Doc Text:	Pacemaker used a 100s timeout for service resources. However, a systemd timeout requires an additional timeout period after the initial timeout to accommodate for a SIGTERM and then a SIGKILL. This fix increases the Pacemaker timeout to 200s to accommodate two full systemd timeout periods. Now the timeout period is enough for systemd to perform a SIGTERM and then a SIGKILL.	Story Points:	---
Clone Of:	1295830	Environment:
Last Closed:	2016-02-18 16:48:14 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Bug Depends On:	1295830
Bug Blocks:	1296673

Description Michele Baldessari 2016-01-05 15:13:19 UTC

+++ This bug was initially created as a clone of Bug #1295830 +++

Via BZ https://bugzilla.redhat.com/show_bug.cgi?id=1275324 we increased the 
stop timeout to 100 seconds.

Initially the 100s recommendation came from the DefaultTimeoutStopSec=90s 
setting in /etc/systemd/system.conf, and I believe the 120s recommendation 
(https://bugzilla.redhat.com/show_bug.cgi?id=1275324#c15) came from
anedoctal evidence observed during test runs, but was lost in the noise of the above BZ.
                                                        
So I took a look at the RHEL 7.2 systemd's source and noticed that the correct formula is actually:
DefaultTimeoutStopSec * 2 + <scheduling-delta*>
                                                                                
* I assume we need a bit of time to make sure that systemd is scheduled, 
that it sends a SIGKILL and that everything (process structures, mainly) is 
gone and five seconds seems quite reasonable (aka if systemd does not get to run within 5 seconds you likely have other issues anyways)
                                                       
This is because in src/core/service.c:static int service_dispatch_timer(sd_event_source *source, usec_t usec, void *userdata) {
...
case SERVICE_STOP_SIGTERM:                                                  
   if (s->kill_context.send_sigkill) {
	   log_unit_warning(UNIT(s)->id, "%s stop-sigterm timed out. Killing.", UNIT(s)->id);
	   service_enter_signal(s, SERVICE_STOP_SIGKILL, SERVICE_FAILURE_TIMEOUT);  
   } else {                                             
	   log_unit_warning(UNIT(s)->id, "%s stop-sigterm timed out.  Skipping SIGKILL.", UNIT(s)->id);
	   service_enter_stop_post(s, SERVICE_FAILURE_TIMEOUT);
   }                            
break;                                                    
...                                                                                                                                    
The man page seems to confirm that systemd will wait one Timout timespan 
for the initial stop request. Then it will send a SIGTERM and wait for another
Timeout to occur and, if the service is still around, then we send a SIGKILL.
"""
TimeoutStopSec=                                                             
   Configures the time to wait for stop. If a service is asked to stop      
   but does not terminate in the specified time, it will be terminated      
   forcibly via SIGTERM, and after another delay of this time with          
   SIGKILL (See KillMode= in systemd.kill(5)). Takes a unit-less value      
   in seconds, or a time span value such as "5min 20s". Pass 0 to           
   disable the timeout logic. Defaults to TimeoutStartSec= in manager       
   configuration file.                                                      
"""

This also confirms that we have seen services still around even after > 100 seconds.

So we need to change the pcs default timeout according to the following formula:
DefaultTimeoutStopSec * 2 + X = 180 + X ~= 185s

This is under the assumption that DefaultTimeoutStopSec in system.conf is
left at the RHEL default of 90 seconds.

Since pacemaker will fence a node when a service fails to stop within the 
configured timeout, this change should avoid most of the spurious fencing 
events when a service was still around after the old 100 seconds timeout.

Comment 2 James Slagle 2016-01-13 21:02:52 UTC

*** Bug 1291474 has been marked as a duplicate of this bug. ***

Comment 4 Ofer Blaut 2016-01-24 07:59:27 UTC

Hi

I can see all openstack services are set to start/stop 200sec
beside Rabbitmq/redis is that by design ? 


Ofer 

[root@overcloud-controller-2 ~]# pcs resource --full  | grep stop -C 1
  Operations: start interval=0s timeout=20s (ip-192.0.2.12-start-interval-0s)
              stop interval=0s timeout=20s (ip-192.0.2.12-stop-interval-0s)
              monitor interval=10s timeout=20s (ip-192.0.2.12-monitor-interval-10s)
--
   Operations: start interval=0s timeout=200s (haproxy-start-interval-0s)
               stop interval=0s timeout=200s (haproxy-stop-interval-0s)
               monitor interval=60s (haproxy-monitor-interval-60s)
--
   Operations: start interval=0s timeout=120 (galera-start-interval-0s)
               stop interval=0s timeout=120 (galera-stop-interval-0s)
               monitor interval=20 timeout=30 (galera-monitor-interval-20)
--
  Operations: start interval=0s timeout=20s (ip-192.0.2.11-start-interval-0s)
              stop interval=0s timeout=20s (ip-192.0.2.11-stop-interval-0s)
              monitor interval=10s timeout=20s (ip-192.0.2.11-monitor-interval-10s)
--
   Operations: start interval=0s timeout=120 (redis-start-interval-0s)
               stop interval=0s timeout=120 (redis-stop-interval-0s)
               monitor interval=45 timeout=60 (redis-monitor-interval-45)
--
   Operations: start interval=0s timeout=370s (mongod-start-interval-0s)
               stop interval=0s timeout=200s (mongod-stop-interval-0s)
               monitor interval=60s (mongod-monitor-interval-60s)
--
   Operations: start interval=0s timeout=100 (rabbitmq-start-interval-0s)
               stop interval=0s timeout=90 (rabbitmq-stop-interval-0s)
               monitor interval=10 timeout=40 (rabbitmq-monitor-interval-10)
--
   Operations: start interval=0s timeout=200s (memcached-start-interval-0s)
               stop interval=0s timeout=200s (memcached-stop-interval-0s)
               monitor interval=60s (memcached-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-nova-scheduler-start-interval-0s)
               stop interval=0s timeout=200s (openstack-nova-scheduler-stop-interval-0s)
               monitor interval=60s start-delay=10s (openstack-nova-scheduler-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (neutron-l3-agent-start-interval-0s)
               stop interval=0s timeout=200s (neutron-l3-agent-stop-interval-0s)
               monitor interval=60s (neutron-l3-agent-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-ceilometer-alarm-notifier-start-interval-0s)
               stop interval=0s timeout=200s (openstack-ceilometer-alarm-notifier-stop-interval-0s)
               monitor interval=60s (openstack-ceilometer-alarm-notifier-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-heat-engine-start-interval-0s)
               stop interval=0s timeout=200s (openstack-heat-engine-stop-interval-0s)
               monitor interval=60s (openstack-heat-engine-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-ceilometer-api-start-interval-0s)
               stop interval=0s timeout=200s (openstack-ceilometer-api-stop-interval-0s)
               monitor interval=60s (openstack-ceilometer-api-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (neutron-metadata-agent-start-interval-0s)
               stop interval=0s timeout=200s (neutron-metadata-agent-stop-interval-0s)
               monitor interval=60s (neutron-metadata-agent-monitor-interval-60s)
--
   Operations: start interval=0s timeout=40 (neutron-ovs-cleanup-start-interval-0s)
               stop interval=0s timeout=300 (neutron-ovs-cleanup-stop-interval-0s)
               monitor interval=10 timeout=20 (neutron-ovs-cleanup-monitor-interval-10)
--
   Operations: start interval=0s timeout=40 (neutron-netns-cleanup-start-interval-0s)
               stop interval=0s timeout=300 (neutron-netns-cleanup-stop-interval-0s)
               monitor interval=10 timeout=20 (neutron-netns-cleanup-monitor-interval-10)
--
   Operations: start interval=0s timeout=200s (openstack-heat-api-start-interval-0s)
               stop interval=0s timeout=200s (openstack-heat-api-stop-interval-0s)
               monitor interval=60s (openstack-heat-api-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-cinder-scheduler-start-interval-0s)
               stop interval=0s timeout=200s (openstack-cinder-scheduler-stop-interval-0s)
               monitor interval=60s (openstack-cinder-scheduler-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-nova-api-start-interval-0s)
               stop interval=0s timeout=200s (openstack-nova-api-stop-interval-0s)
               monitor interval=60s start-delay=10s (openstack-nova-api-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-heat-api-cloudwatch-start-interval-0s)
               stop interval=0s timeout=200s (openstack-heat-api-cloudwatch-stop-interval-0s)
               monitor interval=60s (openstack-heat-api-cloudwatch-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-ceilometer-collector-start-interval-0s)
               stop interval=0s timeout=200s (openstack-ceilometer-collector-stop-interval-0s)
               monitor interval=60s (openstack-ceilometer-collector-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-keystone-start-interval-0s)
               stop interval=0s timeout=200s (openstack-keystone-stop-interval-0s)
               monitor interval=60s (openstack-keystone-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-nova-consoleauth-start-interval-0s)
               stop interval=0s timeout=200s (openstack-nova-consoleauth-stop-interval-0s)
               monitor interval=60s start-delay=10s (openstack-nova-consoleauth-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-glance-registry-start-interval-0s)
               stop interval=0s timeout=200s (openstack-glance-registry-stop-interval-0s)
               monitor interval=60s (openstack-glance-registry-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-ceilometer-notification-start-interval-0s)
               stop interval=0s timeout=200s (openstack-ceilometer-notification-stop-interval-0s)
               monitor interval=60s (openstack-ceilometer-notification-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-cinder-api-start-interval-0s)
               stop interval=0s timeout=200s (openstack-cinder-api-stop-interval-0s)
               monitor interval=60s (openstack-cinder-api-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (neutron-dhcp-agent-start-interval-0s)
               stop interval=0s timeout=200s (neutron-dhcp-agent-stop-interval-0s)
               monitor interval=60s (neutron-dhcp-agent-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-glance-api-start-interval-0s)
               stop interval=0s timeout=200s (openstack-glance-api-stop-interval-0s)
               monitor interval=60s (openstack-glance-api-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (neutron-openvswitch-agent-start-interval-0s)
               stop interval=0s timeout=200s (neutron-openvswitch-agent-stop-interval-0s)
               monitor interval=60s (neutron-openvswitch-agent-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-nova-novncproxy-start-interval-0s)
               stop interval=0s timeout=200s (openstack-nova-novncproxy-stop-interval-0s)
               monitor interval=60s start-delay=10s (openstack-nova-novncproxy-monitor-interval-60s)
--
   Operations: start interval=0s timeout=30 (delay-start-interval-0s)
               stop interval=0s timeout=30 (delay-stop-interval-0s)
               monitor interval=10 timeout=30 (delay-monitor-interval-10)
--
   Operations: start interval=0s timeout=200s (neutron-server-start-interval-0s)
               stop interval=0s timeout=200s (neutron-server-stop-interval-0s)
               monitor interval=60s (neutron-server-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (httpd-start-interval-0s)
               stop interval=0s timeout=200s (httpd-stop-interval-0s)
               monitor interval=60s (httpd-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-ceilometer-central-start-interval-0s)
               stop interval=0s timeout=200s (openstack-ceilometer-central-stop-interval-0s)
               monitor interval=60s (openstack-ceilometer-central-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-ceilometer-alarm-evaluator-start-interval-0s)
               stop interval=0s timeout=200s (openstack-ceilometer-alarm-evaluator-stop-interval-0s)
               monitor interval=60s (openstack-ceilometer-alarm-evaluator-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-heat-api-cfn-start-interval-0s)
               stop interval=0s timeout=200s (openstack-heat-api-cfn-stop-interval-0s)
               monitor interval=60s (openstack-heat-api-cfn-monitor-interval-60s)
--
  Operations: start interval=0s timeout=200s (openstack-cinder-volume-start-interval-0s)
              stop interval=0s timeout=200s (openstack-cinder-volume-stop-interval-0s)
              monitor interval=60s (openstack-cinder-volume-monitor-interval-60s)
--
   Operations: start interval=0s timeout=200s (openstack-nova-conductor-start-interval-0s)
               stop interval=0s timeout=200s (openstack-nova-conductor-stop-interval-0s)
               monitor interval=60s start-delay=10s (openstack-nova-conductor-monitor-interval-60s)

Comment 5 Raoul Scarazzini 2016-01-29 11:30:33 UTC

Hi Ofer,
since Rabbit and Redis are not systemd resources, then it's ok to have different timeout for those, this bug is limited to the systemd resources.

Comment 6 Michele Baldessari 2016-01-29 12:41:39 UTC

Correct, the 200s comes from systemd default timeout (90s) times 2 + delta:
90s*2 + 20s = 200s

Comment 8 errata-xmlrpc 2016-02-18 16:48:14 UTC

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHBA-2016-0264.html

Comment 9 Andrew Beekhof 2016-05-09 04:33:32 UTC

*** Bug 1322387 has been marked as a duplicate of this bug. ***