Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 364734

Summary:

[CI] Jetty server shutdown stalls

Product:

[RT] Virgo

Reporter:

Glyn Normington <glyn.normington>

Component:

jetty

Assignee:

Glyn Normington <glyn.normington>

Status:

CLOSED FIXED

QA Contact:

Severity:

normal

Priority:

CC:

eclipse

Version:

unspecified

Target Milestone:

---

Hardware:

OS:

Mac OS X - Carbon (unsup.)

Whiteboard:

Attachments:

Description	Flags
Full trace log	none

Description Glyn Normington

2011-11-24 11:56:47 EST

Created attachment 207490 [details]
Full trace log

The Virgo Jetty Server CI build 466 failed and was still failiing at build 469: https://hudson.eclipse.org/hudson/view/Virgo/job/virgo.jetty-server.snapshot/469/

It appears that the smoke test cannot complete because VJS is stalled during shutdown, which prevents the next instance from starting.

The full trace log is attached.

The event log shows:
...
[2011-11-24 11:25:36.320] fs-watcher                   <DE0004I> Starting bundle 'org.eclipse.virgo.apps.admin.web' version '3.1.0.D-20111110170746'. 
[2011-11-24 11:25:36.323] start-signalling-1           <DE0005I> Started bundle 'org.eclipse.virgo.apps.admin.web' version '3.1.0.D-20111110170746'. 
[2011-11-24 11:25:36.705] start-signalling-1           <DE0005I> Started bundle 'org.eclipse.virgo.apps.admin.core' version '3.1.0.D-20111110170746'. 
[2011-11-24 11:25:36.707] start-signalling-1           <DE0005I> Started plan 'org.eclipse.virgo.apps.admin.plan' version '3.0.0'. 
[2011-11-24 11:25:40.872] System Bundle Shutdown       <KE0010I> Shutdown initiated.

The trace log shows:
...
[2011-11-24 11:25:42.224] INFO  iLogServiceListener@5e628704 org.osgi.service.log.LogService                                   Bundle org.eclipse.virgo.medic.core_3.1.0.D-20111031170115, Service 63, ServiceEvent UNREGISTERING 
[2011-11-24 11:25:42.224] INFO  iLogServiceListener@5e628704 org.osgi.service.log.LogService                                   Bundle org.eclipse.virgo.medic.core_3.1.0.D-20111031170115, Service 47, ServiceEvent UNREGISTERING 
[2011-11-24 11:25:42.225] INFO  iLogServiceListener@5e628704 org.osgi.service.log.LogService                                   Bundle org.eclipse.virgo.medic.core_3.1.0.D-20111031170115, Service 43, ServiceEvent UNREGISTERING 
[2011-11-24 11:25:42.225] INFO  iLogServiceListener@5e628704 org.osgi.service.log.LogService                                   Bundle org.eclipse.virgo.medic.core_3.1.0.D-20111031170115, Service 39, ServiceEvent UNREGISTERING 
[2011-11-24 11:25:42.225] INFO  iLogServiceListener@5e628704 org.osgi.service.log.LogService                                   Bundle org.eclipse.virgo.medic.core_3.1.0.D-20111031170115, Service 41, ServiceEvent UNREGISTERING 
[2011-11-24 11:25:42.225] INFO  iLogServiceListener@5e628704 org.osgi.service.log.LogService                                   Bundle org.eclipse.virgo.medic.core_3.1.0.D-20111031170115, Service 61, ServiceEvent UNREGISTERING

Comment 1 Glyn Normington

2011-11-29 06:28:47 EST

https://hudson.eclipse.org/hudson/job/virgo.jetty-server.snapshot/475/ also failed

Comment 2 Glyn Normington

2011-12-09 08:42:21 EST

Jetty server commit 50421e1d162274b56e063781db39061bb41c3001 increased a sleep to 2 minutes to allow for the first instance of Virgo to shut down before warm starting the next instance. Apparently Jetty is relatively fast to release its 8080 port and so the next server initialisation could have overlapped the previous server coming down, resulting in potential port clashes which might explain why the second server startup mysteriously started shutting down part way through.

Anyway the build is clean now, so let's see if this recurs.