Confluence: word export takes too long

[IT Einkauf]

Hi!

We use a few confluence instances in the company. In one of them it takes a very long time when you want to execute a word export. No matter how long the article is it takes over a minute until the save popup appears. In the other confluence instances it works very quickly without any waiting time. I have already used the following articles. The pdf export works now after following the instructions in this article

https://confluence.atlassian.com/confkb/pdf-export-takes-longer-when-running-confluence-with-an-outbound-proxy-856713110.html

https://confluence.atlassian.com/confkb/space-export-page-loads-infinitely-or-very-slow-293831099.html

but unfortunately I can't get any further here with the word export. Can anyone help me here?

Thanks a lot

[James Ponting]

Hi @IT Einkauf

Unfortunately this isn't going to be trivial to investigate.

In terms of next steps, I'd be recommending that you take thread dumps from the instance like below

1. Start taking thread dumps every 10 seconds on a timer with a script
2. Begin the export process
3. When the export completes, save the file
4. Stop the thread dump script

With that data in hand, you can review the thread dumps to see where the time is being spent during the export process. With this you can identify if additional resource may be required, or if there's some other blocker.

I'd also recommend reviewing your Garbage Collection logs in Confluence. The log files for Garbage Collection are stored in the <confluence-install>/logs/ directory, and each file begins with "gc-*".

This is one where you may need to engage with Atlassian Support to progress further, as we're not going to be able to assist you in a particularly interactive way through the community portal (we don't want to be putting your log files out in public for instance).

Hopefully this helps.

[James Ponting]

With regards to the thread dumps, if you'll need your Systems Administrator access to the server Confluence is running on, and assist with capturing and reviewing the thread dumps.

If you're running Windows, you can use the following guide https://confluence.atlassian.com/jirakb/using-process-explorer-to-analyze-jira-application-performance-on-windows-707624983.html.

If you're running Linux, you can use the below steps

1. Identify the Confluence Process ID using ps aux | grep -i java
2. Replace $CONF_PID in the below script with the Process ID from above
   for i in $(seq 6); do top -b -H -p $CONF_PID -n 1 > conf_cpu_usage.`date +%s`.txt; jstack $CONF_PID > conf_threads.`date +%s`.txt; sleep 10; done
3. The above script is configured to run 6 times capturing thread dumps each 10 seconds into the directory the script is run from. To increase the duration, change the seq to a the desired number.

[Suvradip Paul]

Hi @IT Einkauf ,

As you have mentioned word  so you are exporting single pages to .doc file ?  Do these pages have too many images in it? 

Please check this https://confluence.atlassian.com/doc/recognized-system-properties-190430.html and check the value of embedded images attribute.

[IT Einkauf]

Hi,

thanks for your help. No, the export to .doc files needs too long with every pages. It doesn't matter if there are many pictures on the page or only 2 sentences to read.