Confluence Slow to Respond

Kody Ensley August 13, 2015

Dear community,

My team's Confluence instance, while only around 100 users, has a prolific documentation team (working for about 3 years solid). In the last few months we have started to notice decent sluggishness when accessing Confluence and requesting pages. 

I have checked the following:

  1. No antivirus is running on that machine
  2. Multiple users on different OS/web browsers are reporting the same thing
  3. No proxy connection to the confluence server
  4. No obvious errors in <confluence-home>/logs/atlassian-confluence.log (tail -n 5000 atlassian-confluence.log | egrep "warning|error")

 

setenv.sh:

CATALINA_OPTS="$CATALINA_OPTS -Xms1024m -Xmx1024m -XX:MaxPermSize=384m -XX:+UseG1GC"

(I know the MaxPermSize flag is obsolete or not in favor anymore, I have yet to look into that)

On the System Information tab in our Confluence Instance the JAVA VM memory Stats:

Heap Memory 
Used Memory (62 %) Free Memory (38 %)
38 % Free
Maximum Heap Size 1024 MB
Total Allocated Memory 1024 MB
Used Memory 630 MB
Free Allocated Memory 393 MB
Total Free Memory 393 MB
Permanent Generation Memory 
Used Memory (85 %) Free Memory (15 %)
15 % Free
Maximum PermGen 384 MB
Used PermGen 326 MB

 

Any of this look glaringly wrong? Any suggestions on things to increase?

We are currently running Confluence 5.7 and are planning to upgrade to the latest release sometime next week. 

Thanks!

 

3 answers

1 vote
Mike Rathwell
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
August 13, 2015

First thing is that I had 5.7 some time ago and I noticed it was a bit pokey regardless. That said, I have my heap set to 2048MB and I also set Xss=8M (thread space). That last gave me a significant overall boost; nothing definitive... it just felt lighter and quick. There's a LOT going on in a Confluence server and thread space could get tight.

All THAT said... Just don't hesitate to head to 5.8.x. I did that and the difference in performance was dramatic. I have over 500 users with significant activity on a 4vCPU 8GB RAM Windows Server 2012R2 (yes, I have reasons for Windoze rather than Linux; some of them good ones) and it just flies. Page redraws are essentially instantaneous. Insane fast. I suspect a lot of THAT comes from move to Java 8 and Tomcat 8. 

Karalee August 13, 2015

We were talking about a similar issue at our Melbourne AUG and we've found that going to Java 8 made a big impact on performance improvement.

0 votes
Luiz Maia
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
August 14, 2015

Hi Kody,

Just adding my input to the other comments. 1024 MB of heap size is a pretty low value. I understand the usage seem to be ok from the statistic, but we might have memory usage increased in certain moments.

I'd suggest you set 2048 MB for both Xms and Xmx values.

Thanks!

Kind regards,

Luiz Maia
Atlassian Support

0 votes
Alex Medved _ConfiForms_
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
August 13, 2015

Also, PermGen memory setting is quite low (try increasing it to 512MB)

1024MB as a JVM heap size for Confluence to run seems also low.

What is the server load and server settings (memory, database is on the same machine?)

Suggest an answer

Log in or Sign up to answer
TAGS
AUG Leaders

Atlassian Community Events