Starting Confluence kills Jira and vice versa

[the-noob]

I've just installed Jira (atlassian-jira-6.0.3-x64.bin) and Confluence (atlassian-confluence-5.1.4-x64.bin) following the steps described at https://confluence.atlassian.com/display/ATLAS/Here+Be+Dragons

I've used the default recommended express setup.

Since I've started with JIRA everything went ok, configuring and all, but while configuring Confluence I got to the step of user management and decided to connect with JIRA... and found out JIRA died... I've started JIRA (console /opt/atlassian/jira/bin/start-jira.sh) and Confluence died...

They are the only apps (almost) running on a VPS with 1GB ram (guess that should be enough for 4 users...)

I've kept a tail -f on /opt/atlassian/jira/logs/catalina.out while setting up again with Confluence but nothing showed.

Any ideas how to debug ?

[the-noob · Accepted solution]

I've just upgraded again the VPS to 3GB of RAM, now they both work in parallel with the default settings.
As a recap:

- 1GB ram - either of them works, you're not able to start them in parallel

- 2GB ram - they can be started in parallel but after using one more the other dies

- 3GB ram - they can work in parallel

Usage: 1 user tabbing slowly between them

[Justin Alex Paramanandan]

Okay, first up: I reckon that the memory for both applications is insufficient. JIRA itself would require 1GB (Max: 768MB; Permgen: 256MB), and I reckon that starting up Confluence eventually caused the JRIA instance to run out of memory and dies out. Confluence itself would require a similar amount of memory, so I reckon that 1GB is not gonna cut it.

You can try to increase the physical memory of your VPS to roughly 3GB. Once you got JIRA up and running, try to execute free -m to see how much memory is available for Confluence.

[the-noob]

Hi,

Justin, if I really need 3GB of ram for 2 idle instances sounds like I've made a bad decision to buy them. I've seen mentions of servers running JIRA unde 1GB.

William, they are running on the default ports (8080- jira, 8090- confluence) and I'm sure I've created separate databases for each but in order to make sure, where can I find the db settings?

[Justin Alex Paramanandan]

True to that, mate! Running JIRA alone under 1GB would be fine, and the 1GB memory setup is the recommended setup for most instances; you could run it at a lower specifications setup, but the suggested threshold is catered for performance related issues that might arise as your instance expands. Thus, providing this amount would suffice in the long run.

However, just as a precaution, performance is not only measured by the number of users or issues alone, but also the number of schemes, comments and workflow steps in the entire instance. You can read more about it here: Scaling JIRA.

I reckon what @William mentioned is the usage of JIRA and Confluence on the same database engine (e.g.: MySQL on localhost; port 3306 having two databases). However, I don't think that would cause this issue, since they are both configured on different databases. Since you would like to make sure of this, open up the following files which state the database settings for the instances:

• JIRA: $JIRA_HOME/dbconfig.xml
• Confluence: $CONFLUENCE_INSTALL/conf/server.xml

[Septa Cahyadiputra]

First I would recommend you to try our log scanner which is available in both Confluence and JIRA to see if there are any known issue which causing such issue on your end.

However, since JIRA instance process just gone I would suggest you to see if there are enough memory allocated for your server. You might need to increase your physical memory.

[the-noob]

I'm running a default Ubuntu 12.04.3 LTS, nothing else installed beside JIRA and Confluence.

'stopped working' means the jira process dies and I have to go back in console and use jira/bin/start-jira.sh to start it again

Don't know how to debug thread traces, any hints on that please ?

[Nic Brough -Adaptavist-]

System usage monitoring and the logs are about the only way to find out what's going on here. We don't really have a lot about your configuration, what's running on the machine, what resources are in use. Have you tried debugging thread traces on the one that has "stopped working"? And when you say "stopped working", what does that actually mean?

[the-noob]

I really have to reopen this as it's becoming ridiculous.

I've upgraded the VPS holding only 2 idle instances of JIRA and Confluence to 2GB RAM.
Now I'm able to use them both in parallel but after a while one of them just stops working.

I.E. I'm starting both of them now.. do my work and by tomorron either JIRA or Confluence stops.

How can I debug this ?

[the-noob]

This is a bit disspointing, the minimum system requirements should be stated clearly.

Still it's very strange as the system info shows 50%+ free memory (each type)

[Nic Brough -Adaptavist-]

Mmm, they are mentioned in the docs (e.g. a "small" Jira would run comfortably on most modern off-the-shelf machines with 2Gb RAM), but I certainly would not argue if you described that information along the lines "not obvious".

It's there - https://confluence.atlassian.com/display/JIRA/JIRA+Requirements but I had to hit a search engine to find it, rather than finding links on pages I'd expect people to find if they are new and thinking "let's give this atlassian stuff a go"

[Renjith Pillai]

@the-noob

It depends. If you plan to run JIRA, Confluence and the two databases on a 1 GB instance, it may be really tough for you to manage. Remember we are talking about a server instance serving two web servers with 1GB. You may add some swap space, but the instances will be extrememly slow, making it practically useless. JIRA alone is still okay.

[William Zanchet [Atlassian]]

Hey, you can check if they're running under the same port in their respectives server.xml files. Also, be sure that Confluence and JIRA, are not using the database. We don't recommend to use shared databases.