Details
-
Improvement
-
Status: Resolved
-
Normal
-
Resolution: Fixed
-
None
-
None
-
None
-
-
Night's Watch
-
3
-
NW - 2021-02-03, NW - 2021-02-17
-
Not Needed
-
Needs Assessment
Description
Description
Out of the 4059 CI runs from the last 7 days , 78 failed to install an agent. That's 1.9% failure rate, which is a big chunk of the environmental failures we see in modules testing.
Looking through the errors, most are from RHEL/yum failing to run properly, followed by a flurry of issues - from timeouts to fileaccess denied - on windows.
See the attached agent_install_failed.json file for a list of all errors.
Expected Behaviour
puppet_agent::install task doesn't fail on insulated network or windows hiccups.
Suggested Mitigation Strategy
The puppet_agent::install task retries downloading/installing the agent in the face of transient network or windows errors to avoid spurious environmental errors becoming customer-visible faults.