Timeout : small docx to pdf transformation / conversion

[oshikaru]

Hello !

In production we had an issue last week, Alfresco was not able to generate a PDF preview of a docx document.

We had multiple cases that could explain this (document size, status of soffice, permissions issues) but none of them can explain this behavior today.

This issue is reproductible for this document only on all our environnements (running Community - 5.2.0 (re21f2be5-b22)).

The document have nothing in particular that I can see, it's like all our others documents (I mean thousands of documents) and surely not the heavier.

Here are the only understandable output I can find in the logs : 

11:03:20,293 DEBUG [org.alfresco.repo.jscript.RhinoScriptProcessor.calls] [http-bio-8080-exec-10] thumbnail.get.js Exception
org.mozilla.javascript.WrappedException: Wrapped org.alfresco.service.cmr.thumbnail.ThumbnailException: 10250010 10250009 Transformation failed to obey timeout limit (classpath*:alfresco/templates/webscripts/org/alfresco/repository/thumbnail/thumbnail.get.js#80)

Alfresco is configured to use libreoffice (soffice process) to generate it's previews and it works like a charm using the libreoffice command line so that's why I'm not posting this request in Libreoffice bug tracker.

I know it will be a bit tricky to find a solution with this small report but maybe you could have some keys to help me to debug this error.

Thanks in advance !

[oshikaru]

After more analysis (and a tcpdump between Alfresco and LO.. Ugh), the issue is definitly a communication issue between Alfresco and LO.

So Alfresco is giving the document to onlyoffice, creating the target document but seems to don't send the command to generate the preview.

Here is an exemple : 

Target file when the preview works : 

-rw-r--r--. 1 root   root   270343 Nov 29 15:59 OpenOfficeContentTransformer-target-3981768311599418742.pdf

Target file when it doesn't (0k) : 

-rw-r--r--. 1 tomcat tomcat      0 Nov 29 15:59 OpenOfficeContentTransformer-target-8330268925119395651.pdf

Communication when it works (hard to read, you should use https://www.diffchecker.com) : 

https://pastebin.com/SA9Zfqv5

And when it doesn't :

https://pastebin.com/zga7dvrt

The communication seems to end before all operations are completed (the highlighted part) : 

https://ibb.co/QkHRC89

I will open an issue on Jira but if you have any clue about this please let me know.

Thanks for your time !