Bug 889364 - Site fails to load after git push and there is a confusing message about starting the application
Summary: Site fails to load after git push and there is a confusing message about star...
Keywords:
Status: CLOSED UPSTREAM
Alias: None
Product: OpenShift Online
Classification: Red Hat
Component: Containers
Version: 1.x
Hardware: i686
OS: Linux
low
low
Target Milestone: ---
: ---
Assignee: Rob Millner
QA Contact: libra bugs
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2012-12-20 22:59 UTC by Matt
Modified: 2013-11-18 00:45 UTC (History)
6 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2013-05-17 23:32:17 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)
output of the git push (3.78 KB, application/octet-stream)
2012-12-20 22:59 UTC, Matt
no flags Details

Description Matt 2012-12-20 22:59:05 UTC
Created attachment 666981 [details]
output of the git push

Description of problem:
Application http://flirtinator-ferociousflirt.rhcloud.com/ reports The page isn't redirecting properly on Firefox

Also, there is a message in the output of git push that is confusing:

remote: Application is explicitly stopped!  Use 'rhc app start -a jbossas-7' to start back up.

How reproducible:

Make a small change in the local files and do a git push

  
Actual results:

It appears to deploy fine except for the message to start the app and also that the appears is not there

Expected results:

The app to appear

Additional info:

Comment 1 Nam Duong 2012-12-20 23:04:13 UTC
To add, looks like the application is idled and is failing to restart on git push.  

I tried reproducing with php-5.3 app but wasn't able to, so it could be specific to Jbossas-7?  

NOTE:  We're idling apps after 2 days.  This one doesn't seem to start up on first http request.  

Here's IRC discussion:  

[14:25]  <m0smith> When I do a git push, this is one of the lines of output
[14:25]  <m0smith> remote: Application is explicitly stopped!  Use 'rhc app start -a jbossas-7' to start back up.
[14:27]  <m0smith> rhc app show flirtinator --state
[14:27]  <m0smith> RESULT:
[14:27]  <m0smith> Geargroup jbossas-7 is idle
[14:28]  <m0smith> even thought rhc app start flirtinator  returns started

Comment 2 joycezhang 2012-12-21 07:25:36 UTC
Please refer to the details steps for reproducing this bug as below:

Build:
devenv_2618
rhc-1.3.2.gem

Steps:
1. Create a jbossas app
#rhc app create jbossas1 jbossas-7 -p xx
2. idle this app on instance
#oo-idler -u uuid
3. Do some changes in local repo and git push
4. Try to visit this app via Browser

Actual results:
To step3: It's deployed fine but the final message is confusing as mentioned above.
remote: Running .openshift/action_hooks/build
remote: Running .openshift/action_hooks/deploy
remote: hot_deploy_added=false
remote: Application is explicitly stopped!  Use 'rhc app start -a jbossas-7' to start back up.
remote: Done
remote: Running .openshift/action_hooks/post_deploy
To ssh://189aaed28993492dba6997b8a0d7c188.rhcloud.com/~/git/jbosas1.git/
   17ac4f6..01fe695  master -> master

To step4: The app can be redirected correctly, the changes take effect on the weg page. And the app status is changed to running automatically in this case.

Comment 3 Dan Mace 2013-01-03 20:59:12 UTC
Downgrading the severity.

The message is indeed confusing, as it doesn't distinguish between user-stopped and idler-stopped states, even though the platform has the information to make such a distinction in the output.

This will be addressed in a future release when the relevant cartridge start logic is pulled up into the platform and taken out of the cartridge scripts.

Comment 4 Dan Mace 2013-01-03 21:04:02 UTC
A quick addendum: Building an idled app won't restart it, by design. I suspect that the request made which apparently didn't restore the app right away was a dropped request at the beginning of the restore cycle before JBoss was capturing requests.

Additionally, the "state" of the application being started in that case is expected. The state is set optimistically prior to actually starting the application, and represents the intended state. So, there may be a brief window of time where the app is "started" but still not servicing requests while it's actually starting up.

Comment 5 Rob Millner 2013-05-17 23:32:17 UTC
Feature added to upstream.  The new cartridge API starts idled apps on build.


Note You need to log in before you can comment on or make changes to this bug.