Bug 1366581

Summary:	[NSAPI] Redirecting makes web server crash
Product:	[JBoss] JBoss Enterprise Application Platform 6	Reporter:	Jan Stefl <jstefl>
Component:	Apache Server (httpd) and Connectors	Assignee:	Mladen Turk <mturk>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Jan Stefl <jstefl>
Severity:	urgent	Docs Contact:
Priority:	unspecified
Version:	6.4.9	CC:	jclere, jstefl, mbabacek, mturk, rhatlapa
Target Milestone:	---
Target Release:	EAP 6.4.9
Hardware:	Unspecified
OS:	Solaris
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:
Clones:	1369758 (view as bug list)		Environment:
Last Closed:	2017-01-17 14:47:59 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	1369758

Comment 1 Mladen Turk 2016-08-18 14:00:00 UTC

I would try to use a different shm_file inside magnus.conf
e.g.
.... shm_file="/tmp/eap649-native/jk_shm" 
.... shm_file="/tmp/jbcs246-native/jk_shm" 

In case of failure shared memory on Solaris can get locked until system reboot

Comment 2 Jan Stefl 2016-08-18 16:03:44 UTC

Thank you for the hint! I have updated tests and started them. I will know more tomorrow.

Comment 3 Jan Stefl 2016-08-19 14:50:53 UTC

I have changed the folder and let the machine restart.
Unfortunately, problem still exists.

I have notice, there is no file on /tmp/eap649-native/jk_shm, I would expect it there because was configured in magnus.conf. The same problem for /tmp/eap649-native/nsapi.log




-------
iPlanet log
-------
finest ( 4780): for host 10.16.92.95 trying to GET /nc1/SysProperties.jsp, func_exec reports: fn="force-type" type="text/plain" Directive="ObjectType" returned 0 (REQ_PROCEED)
[19/Aug/2016:10:43:09] finest ( 4780): for host 10.16.92.95 trying to GET /nc1/SysProperties.jsp, func_exec reports: executing fn="jk_service" worker="worker1" path="/nc1(/*)" Directive="Service"
[19/Aug/2016:10:43:09] catastrophe ( 4780): CORE3260: Server crash detected (signal SIGSEGV)
[19/Aug/2016:10:43:09] info ( 4780): CORE3261: Crash occurred in NSAPI SAF jk_service
[19/Aug/2016:10:43:09] info ( 4780): CORE3262: Crash occurred in function jk_shm_check_maintain from module /mnt/hudson_workspace/workspace/eap-60-native-connectors-solaris-sparc/jdk/java16_default/label_exp/dev33-solaris11-sparc/jbosseap/modules/system/layers/base/native/lib64/nsapi_redirector.so
[19/Aug/2016:10:43:44] fine ( 4779): Child process closed admin channel
[19/Aug/2016:10:43:44] fine ( 4779): CORE3061: received signal 18
[19/Aug/2016:10:43:44] fine ( 4779): CORE3049: Detected child process 4780 died (status 139)
[19/Aug/2016:10:43:44] info ( 4779): CORE3050: Spawning replacement for worker process 4780
[19/Aug/2016:10:43:44] fine ( 4779): Marked slot for pid = 4780 as empty

Comment 5 Mladen Turk 2016-08-23 08:51:24 UTC

Hmm, could you post the complete magnus.conf for the new changed path setup?

Perhaps all paths should be the same.
Anyhow, if not inside configured location, where is nsapi.log created ?

Also can you post nsapi.log (the one from mpod_jk)
By trace it looks like problem originates from shared memory, so if the same segment from previous version is opened it will have different layout and size, causing problems.

Comment 6 Jean-frederic Clere 2016-08-23 09:44:00 UTC

Please retry with nsapi_redirector.so from http://www.qa.jboss.com/xbuildroot/packages/ep6/mod_jk/1.2.41/2.sun10/ and tell us if that helps

Comment 7 Jan Stefl 2016-08-23 11:40:24 UTC

I can confirm that problem was fixed.
Thank you!

Comment 9 Jan Stefl 2016-09-20 09:00:17 UTC

Issue was fixed

Tested on EAP 6.4.10/natives based on EWS 2.1.1.

Comment 10 Petr Penicka 2017-01-17 14:47:59 UTC

Retroactively bulk-closing issues from released EAP 6.4 cumulative patches.