Jira 4.4.3 Crashes shortly after install on Ubuntu 11.10

[hobsie]

I have had a look around the questions and answers but couldn't find an exact match to the issues I'm having.

I'm trying to install Jira 4.4.3 on a basic package Linode cloud server box running Ubuntu server 11.10 with 512MB Ram, 20GB HD space (not sure about processing speeds).

After installing the standalone .bin package I go to the web wizard and from here it starts to go a little wrong. Most of the time I can make it to the end of the wizard (I have tried both with an internal database and external MySQL) right up to the point of suggesting I log in. However sometimes during this process (and 100% of the time when i get to the end) the Jira process seems to crash (or at least it drops off the "top" process list) and whenever I try start jira with ./jira-start.sh the same will eventually happen after a couple of minutes.

I've had a look through the logs in /var/atlassian/applweb116-jira2ication-data/jira/log/*.log and nothing really jumps out at me, just sort of stops. My first instinct was that perhaps I was simply running out of memory but I have tried adjusting the Jira memory heap use and even if I don't do that "tops" shows that Jira never really gets above 70% of memory (there's sweet f all else running during this time).

The next idea I had was that maybe it's something to do with the version of Java as I've seen a few warnings about the version of java shipping with ubuntu (and I beleive java6 was deprecated on ubuntu recently) but even after getting sun-java6-jdk and plugin installed I get the same issue.

Anyway, at the moment I'm at a bit of a loss as to ways for me to find out what the issue is. The logs haven't been very helpful but perhaps I'm just looking in the wrong place or even in the wrong logs?

Regards

Hobsie

Edit: So I have managed to get a little further now. I found that my $JAVA_HOME had somehow gotten unset (probably while getting sun-java6-jdk as it doesn't come with ubuntu 11.10) so I fixed that. I can now log in to the server but, again, after a short period of time the server 'crashes' (process disapears).

Last bit of catalina.out looks like this:

2011-11-09 16:30:40,354 main INFO [atlassian.jira.startup.JiraStartupLogger]

********************************************************************************

JIRA 4.4.3 build: 663 started. You can now access JIRA through your web browser.

********************************************************************************

2011-11-09 16:30:52,829 main INFO [atlassian.jira.scheduler.JiraSchedulerLauncher] Starting the JIRA Scheduler....

2011-11-09 16:30:53,031 main INFO [atlassian.jira.scheduler.JiraSchedulerLauncher] JIRA Scheduler started.

Nov 9, 2011 4:30:55 PM org.apache.coyote.http11.Http11Protocol start

INFO: Starting Coyote HTTP/1.1 on http-8080

Nov 9, 2011 4:30:55 PM org.apache.catalina.startup.Catalina start

INFO: Server startup in 83908 ms

2011-11-09 16:31:01,981 QuartzWorker-0 INFO ServiceRunner Backup Service [jira.bc.dataimport.DefaultExportService] Data export completed in 8443ms. Wrote 575 entities to export in memory.

2011-11-09 16:31:02,172 QuartzWorker-0 INFO ServiceRunner Backup Service [jira.bc.dataimport.DefaultExportService] Attempting to save the Active Objects Backup

2011-11-09 16:31:02,858 QuartzWorker-0 INFO ServiceRunner Backup Service [jira.bc.dataimport.DefaultExportService] Finished saving the Active Objects Backup

2011-11-09 16:31:06,040 Modification Check:thread-1 INFO [atlassian.jira.startup.JiraStartupLogger]

___ Modifications ___________________________

Modified Files : jira-application.properties

Removed Files : None

[Kinto Soft · Accepted solution]

I use JIRA 4.4.3 on a Linode (Ubuntu 10.01) with 1.5 Gb RAM and every week I need to reboot the system because it hungs.

It's for demo, and there are is only one project with less than 5 issues and likely about 5-10 visitors/day.

I asked Linode Support for a solution and they analyzed the system logs without success and they never provided me a good solution.

Would be interesting to know more cases using JIRA hosted on ISPs. Perhaps this affects to Linode only or maybe is a more general problem.

[hobsie]

I have basically come to the conclusion that I cannot get Jira to run on my Linode 512MB in a stable enough fashion.

I haven't been able to truly assertain why this is the case but my gut feeling leads me to believe that it is the Java runtime that is crashing. This may simply be an issue with the server itself and not Jira.

Thanks to those that tried to help.

Hobsie

[JamieA]

There should also be a catalina.out file in there somewhere... (do a find). If the jvm is crashing it will dump the stack trace to that file. Also there is a jvm argument to make it dump core on crash that you could add, if you're familiar withgdb that could help.

> or at least it drops off the "top" process list

I'm not sure whether you mean it's not in the top 20 busiest processes, or it no longer exists. If not then try the above.

[hobsie]

Thanks for the response. When I said drops of the "top" process list I did indeed mean that the process is no longer in the list at all. I'm afraid I'm not at all familiar with gdb (it's all been visual studio for me up to this point). I've taken a look at the catalina.out and the catalina.yyyy-mm-dd.log as well but again nothing very revealing. The .log last entry is:

Nov 8, 2011 10:38:44 PM org.apache.catalina.startup.Catalina start

INFO: Server startup in 48072 ms

And the last entry in catalina.out is:

2011-11-08 22:43:21,633 http-8080-6 INFO anonymous 1359x37x1 y9r25v xxx.xxx.xxx.xxx /secure/SetupDatabase.jspa [atlassian.plugin.loaders.ScanningPluginLoader] Plugin Unit: /var/atlassian/application-data/jira/plugins/.bundled-plugins/workflow-designer-plugin-2.0.0.jar (1318871786000) created

(I've stripped my ip address out). The file is pretty large, too large to paste here, but having a search for keywords like error doesn't come up with much either. This time it crashed in the middle of setting up the external MySQL database, but like I say I normally make it to the end of the wizard before it dies a death.

Regards

Hobsie

[Valentijn Scholten]

I have been having similar issues, but on Windows Server 2003/2008.

Seems that since jira 4.3 we're having out of memory issues.

It's not a heap space or permgen, but an unspecified outofmemory error.

Out JIRAxxxxxxstdderr.log looked like this:

15-nov-2011 14:29:53 org.apache.catalina.core.StandardService start
INFO: Starting service Catalina
15-nov-2011 14:29:53 org.apache.catalina.core.StandardEngine start
INFO: Starting Servlet Engine: Apache Tomcat/6.0.32
java.lang.OutOfMemoryError

Symptoms are excatly like the ones you specify.

Despite the fact that according to atlassian jira doesn't need so much memory, the only way we got around this issue is to install and configure a 64-bits JDK/JRE and set memory to something over 1024M.