Details
-
Bug
-
Status: Resolved
-
Normal
-
Resolution: Fixed
-
None
-
None
-
Coremunity
-
Platform Core KANBAN
-
Needs Assessment
-
Bug Fix
-
Fixes a regression in 6.14.0 that caused puppet agents to process their server_list up to 4 times per agent run, instead of once per run. This restores the behavior that existed in 6.13.0 and before.
-
Needs Assessment
Description
In 6.13.0, if puppet is configured to use server_list, it tries to connect to each entry in the list, stopping at the first HTTP 200 connection to puppetserver's simple status endpoint, and adds the resolved server and port into puppet's context, which is then looked up each time a connection is made. Each time connection attempt may result in a 2 minute timeout for each failed connection. The timeout defaults to 2 minutes, but is configurable using the http_connect_timeout puppet setting.
In 6.14.0, puppet will resolve server_list for puppet, fileserver and report services. This can result in additional delay during the run. So if 6.13.0 was delayed X minutes, then puppet 6.14.0 will be delayed 3X minutes. The exact number depends on how many entries in server_list fail before a successfully connection is made.
Attachments
Issue Links
- causes
-
PUP-10844 Agent failures with server_list when one puppetserver fails
-
- Closed
-