Skip to content

Conversation

default50
Copy link
Contributor

@default50 default50 commented Apr 27, 2016

The agent would fail to start if by chance the (stale) pidfile would reference an existing process with only codedeploy in its name string.

This is not that odd considering:

  • The agent is set up to start at boot.
  • That the stop process takes long enough so that when an instance is shut down it always leaves the pidfile behind (see related issue codedeploy-agent stop service #55)
  • The "@reboot" cronjob is sleeping for 45 seconds before executing /opt/codedeploy-agent/bin/install wich contains codedeploy in the name of the process.

Add enough instances to the mix and you get this issue pretty often.

Just by checking for a more specific string in cmdline this can be avoided.

@coveralls
Copy link

Coverage Status

Coverage remained the same at 94.557% when pulling ea39cb8 on default50:process_matcher_fix into 65ace85 on aws:master.

@yubangxi yubangxi merged commit b9862ae into aws:master May 4, 2016
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants