Issue #440
Restarting Pulp after changing concurrency left stale workers
Description
Description of problem:
Change PULP_CONCURRENCY in /etc/default/pulp_workers prior to restarting pulp_resource_manager, pulp_workers, pulp_celerybeat has left several stale workers on my machine.
Version-Release number of selected component (if applicable):
mod_wsgi-3.4-1.pulp.el6sat.x86_64
pulp-puppet-tools-2.4.0-0.18.beta.el6sat.noarch
m2crypto-0.21.1.pulp-10.el6sat.x86_64
pulp-rpm-plugins-2.4.0-0.18.beta.el6sat.noarch
pulp-katello-plugins-0.3-1.el6sat.noarch
python-pulp-rpm-common-2.4.0-0.18.beta.el6sat.noarch
python-pulp-bindings-2.4.0-0.18.beta.el6sat.noarch
createrepo-0.9.9-21.2.pulp.el6sat.noarch
python-pulp-common-2.4.0-0.18.beta.el6sat.noarch
pulp-server-2.4.0-0.18.beta.el6sat.noarch
pulp-nodes-parent-2.4.0-0.18.beta.el6sat.noarch
pulp-nodes-common-2.4.0-0.18.beta.el6sat.noarch
pulp-selinux-2.4.0-0.18.beta.el6sat.noarch
python-isodate-0.5.0-1.pulp.el6sat.noarch
python-pulp-puppet-common-2.4.0-0.18.beta.el6sat.noarch
python-kombu-3.0.15-5.pulp.el6sat.noarch
pulp-puppet-plugins-2.4.0-0.18.beta.el6sat.noarch
How reproducible:
Always
Steps to Reproduce:
1. change PULP_CONCURRENCY to < available cores on machine in /etc/default/pulp_workers while pulp is running
2. service pulp_resource_manager restart; service pulp_workers restart; service pulp_celerybeat restart;
3. view processes to see if stale processes remain
Actual results:
Pulp only stops and restarts the number of workers as set to PULP_CONCURRENCY.
Expected results:
Pulp should stop all Pulp workers and start only the number specified in PULP_CONCURRENCY or if the value is commented out then the default.
Additional info:
+ This bug was cloned from Bugzilla Bug #1102369 +
History
#1
Updated by dgregor@redhat.com over 3 years ago
Just ran into this myself as well.
+ This comment was cloned from Bugzilla #1102369 comment 1 +
#2
Updated by bmbouter over 3 years ago
- Severity changed from Low to 1. Low
#3
Updated by bmbouter almost 3 years ago
A valid workaround to this issue is to:
1) stop pulp services
2) modify PULP_CONCURRENCY
3) start pulp processes
This will leave no stale processes because they were all stopped in step (1)
Please register to edit this issue