Details
-
Bug
-
Status: Closed
-
Normal
-
Resolution: Fixed
-
None
-
None
-
Bolt
-
Bolt Kanban
-
Customer Feedback
-
Normal
-
1 - 1-5% of Customers
-
2 - Annoyance
-
4 - $$$$$
-
Unclear the impact of this beyond two running processes. More information needed on the impact of this. Does the broker flap between them? Do commands get executed twice? How does it end up without the PIDfile in the first place?
-
34307
-
1
-
Bug Fix
-
This modifies the pxp-agent service to kill all pxp-agent processes when the service is restarted, rather than just the current pxp-agent process.
Description
OS / Version: Confirmed in Centos 6.7 and SLES 11
Reproduction:
/opt/puppetlabs/puppet/bin/pxp-agent --version
|
1.8.1 |
|
#pxp-agent process running normally:
|
|
service pxp-agent status
|
pxp-agent (pid 16099) is running... |
|
rm /var/run/puppetlabs/pxp-agent.pid
|
|
service pxp-agent start
|
Starting PXP agent: [ OK ]
|
|
ps aux | grep pxp-agent | grep -v grep
|
root 16099 1.6 0 1 379536 9776 ? Sl 21:21 0:02 /opt/puppetlabs/puppet/bin/pxp-agent |
root 16297 9.9 0.0 379536 7744 ? Sl 21:23 0:02 /opt/puppetlabs/puppet/bin/pxp-agent |
Expected behavior:
The init script should only allow one running pxp-agent process
Actual behavior:
With a missing pidfile the init script will start another pxp-agent process
Attachments
Issue Links
- relates to
-
PCP-862 pxp-agent connection flapping exhausts resources on broker
-
- Resolved
-
-
MODULES-8533 Windows upgrades are failing due to pxp-agent issues
-
- Closed
-