Hi,
After many Migration Steps of a Plone Site i get the the following Error when i rebuild the Catalog:
ERROR PortalTransforms Cannot register transform html-to-captioned (ImportError), using BrokenTransform: Error
No module named html2captioned
Segmentation fault (core dumped)
Segmentation fault (core dumped)
Segmentation fault (core dumped)
I migrate from Plone 2.5.5 -> 3.3.6 -> 4.0.9 -> 4.3.10 -> 5.0.5. Is it possible that this Transformation is lost by removing the Kupu Tool? What should i do? Or can i ignore this Error?
Allthough the error messages are next to each other I think they are unrelated and your segmentation faults have another cause.
I see those kupu messages often on sites migrated from plone 3 to 4 where a transform has not been unregistered properly, but it's a harmless error message and with a small debug session to update the registration data structure in the transform tool it can be removed.
python segmentation faults are nasty to debug. Do you use the latest python 2.7 version on your system? Do you use any C compiled extensions like database drivers. Those and the python interpreter itself will throw segmentation faults and not a normal 'nice' python traceback.
Someone with gdb experience can help you tracking down what causes the segmentation faults, but it's not easy.
I use the a normal Installer Version of Plone with its own Python. I have copy my Mountpoint with file and blobstorage to the new Installation. I dont use Database Drivers. Hmm...
@1letter Out of curiosity, are you running on an Ubuntu 16.04 machine? There is a Python 2.7 segfault bug on Ubuntu 16.04 that has been reported and is not yet closed. Could your problem be related to that?
You may use strace & ltrace to observe which system/shared libraries calls are made just before the segfaults. I believe that might help checking if it's DB related, for instance.
Ok, i get now a long Error Report from my Ubuntu. It's a Problem with 'wv'. Perhaps are old word-documents the problem, if the catalog reindex the Content of these files.
@1letter I believe that if wv is installed on a machine, Plone will attempt to use it to index the contents of binary files like .doc and .docx. If the binary that does that work is broken, maybe there's a problem that is bringing Plone down as a result. Perhaps you could try updating wv. Failing that, try uninstalling it entirely and see if you can get Plone to stop trying to use it to index things.