[SERVER-15] The puppet server should not die when provided a HUP signal. Created: 2014/09/29  Updated: 2016/02/06  Resolved: 2016/02/06

Status: Closed
Project: Puppet Server
Component/s: None
Affects Version/s: SERVER 0.2.1
Fix Version/s: None

Type: Improvement Priority: Normal
Reporter: Trevor Vaughan Assignee: Unassigned
Resolution: Duplicate Votes: 3
Labels: startup-time
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Any


Issue Links:
Duplicate
is duplicated by SERVER-96 Support HUP for faster restart Closed
Relates
relates to TK-149 Runtime refresh of Jetty CRL (jetty 1.8) Resolved
relates to TK-202 Support HUP signal for restarting ser... Resolved
relates to SERVER-537 Cleanup JRubyPuppet-related resources... Closed
relates to TK-86 validate service config before startup Ready for Engineering
relates to SERVER-396 Revisit AOT in terms of impact on sta... Closed
Template:
Epic Link: HUP support
Sub-team: emerald
QA Contact: Erik Dasher

 Description   

When provided with a HUP signal, the puppet server should not die.

Ideally, it would re-read its configuration files and apply them after cleaning up its external connections. This would prevent the unnecessary interruption of client Puppet runs.

This is currently possible with Passenger + Apache using the 'graceful' httpd restart.



 Comments   
Comment by Chris Price [ 2014/09/30 ]

HUP should result in a graceful shutdown, correct?

We agree that being able to send a signal to cause reloading of config files would be a valuable feature. We'll try to get it on the roadmap!

Eric Sorenson ping

Comment by Eric Sorenson [ 2014/09/30 ]

SIGHUP should be a "reload my configs" signal. Absolutely agree this is an important thing, especially given the lengthy startup time.

Comment by Kevin Corcoran [ 2015/02/04 ]

Brice Figureau - This is the ticket about configuration reloading that I mentioned.

Comment by Chris Price [ 2015/07/14 ]

Thinking that this would probably end up being implemented by just stopping and restarting all of the TK services. We do this in the REPL all the time, so it seems like it might not be that hard to implement; the only tricky part would be to figure out how to register the signal handler, etc. Ping Archana Sridhar this is related to SERVER-373 and TK-149.

Comment by Chris Price [ 2015/07/14 ]

We should probably have a TK ticket that expresses that goal more generally (if we don't already).

Comment by Jeremy Barlow [ 2015/07/15 ]

Archana Sridhar, Chris Price - I linked SERVER-537 to this one. I had been thinking that we would want to complete that one - basically having Puppet Server cleanup resources from the currently running service - in the same Puppet Server release that would include any TK-level work to support config reloading without a process restart.

Comment by Chris Price [ 2015/07/16 ]

Jeremy Barlow good call, thanks for linking.

Comment by Chris Price [ 2016/02/06 ]

We're tracking this work in the epic SERVER-96, so, closing this as a dupe.

Generated at Tue Jan 22 08:20:08 PST 2019 using JIRA 7.7.1#77002-sha1:e75ca93d5574d9409c0630b81c894d9065296414.