upgrade from 5.0.d to 5.1.e: pdfbox error

[vincent-kali]

Hi,
After an Alfresco update from 5.0.d to 5.1.e, everything seems to work fine except
that a pdfbox error prevent from indexing my pdf files:

 ERROR [pdfbox.filter.FlateFilter] [http-bio-8443-exec-2] FlateFilter: stop reading corrupt stream due to a DataFormatException
‍‍‍

Did somebody faced something similar ?
(I've seen that pdfbox fires this kind of message in case of 'out of memory', but server memory is not overloaded in my case)

Thanks for your advise,
Vincent

[talleyrand]

I have installed the latest version today and tried to migrate from 5.0d. I get a lot of this kind of messages and can't preview lots of documents too…

[iblanco]

I'm hanting for an indexing issue that is causing OutOfMemory errors and I'm starting to suspect that the culprit is PDFBox . My instalaltion is alfresco 5.1g. It uses PDFBox-1.8.10 and I found an issue in Tika that suggests that this version might not be a very good one:

 [TIKA-1737] PDFBox 1.8.10 is still a basket case - ASF JIRA.

I made a memory dump and I'm trying to analyze it with Eclipse MAT, the "Leak Suspects" report suggests that 75% of the heap is full of PdfBox's COSObjects that are being retained by a classloader.

Not sure how to interpret this but PDFBox seems to be in the middle.

[mauro1855]

Yes, I also have the same issue.

I had this problem in 5.0 and I continue having it in 5.1.e.

However I don't think there is a pdfbox version (alfresco-patched) more recent than 1.8.10. Meaning we have to continue using this version, right?

I'll try to dig a bit more