Confluence 4.1 - Content Indexing: pdfbox warn/exception during re-index

[Gavin Fowler]

Hi

I've just upgraded to confluence 4.1 (from 4.0.5) and on start-up the content is reindexed. I can see from confluence log the following pdfbox warning/exception:

2012-01-06 10:08:10,383 WARN [scheduler_Worker-8] [apache.pdfbox.util.PDFStreamEngine] processOperator java.io.IOException: Error: expected hex character and not  :32
java.io.IOException: Error: expected hex character and not  :32

Googling the error is seems to be related to the bundled version of pdfbox (pdfbox-1.2.1.jar), and has been reported here: https://jira.atlassian.com/browse/CONF-22358

My question is, should i manually upgrade pdfbox-1.2.1.jar (latest stable version is pdfxbox-1.6.0.jar), or is it best to wait for an official atlassian confluence update?

Many thanks.

Gavin.

[Gavin Fowler · Accepted solution]

Here's the response from Atlassian Support.

Since this does not usually indicate an actual problem, and is more of a warning rather than an error, you should just suppress the messages from being printed in the logs by navigating to <tt>ConfluenceInstallationDir/confluence/WEB-INF/classes/</tt>, edit <tt>log4j.properties</tt> and add this line to the end of the file:

log4j.logger.org.apache.pdfbox.util.PDFStreamEngine=ERROR

Then restart Confluence. We are not aware of issues that you might face if you manually upgrade the pdfbox jar to 1.6, so we will not recommend that for now. Since the issue is already in progress, the fix should be rather soon, so best to wait for that and suppress the warning messages for now.

[David Peterson]

My suggestion would be to contact Atlassian Support and get their recommendation, via http://support.atlassian.com/. I suspect that the pdfbox packages are only used within whatever plugin is responsible for rendering to PDF - it seems a lot more locked down than it used to be.

[Gavin Fowler]

Thanks David. I'll open a support request and post their response.