Failed indexing process with V15.0.504 #88550

https://thebrain.zendesk.com/agent/tickets/99550

Hello Xavier,

Thank you for sending the log file, that was very helpful.

We reviewed it and can see that the indexing process is handling a very large number of attachments, over 5,000 files, many of which appear to be sizable PDFs. The indexing is not completely frozen, but it is progressing extremely slowly and eventually gets marked as “stuck” when no progress is detected for several minutes, which causes the process to stop automatically.

From the log, a few specific PDF files stand out as being particularly heavy during text extraction. It’s possible that one or more of these files is causing the indexing to stall.

Here are a few things that would help us move this forward:

1. If you’re willing, please send us one or both of the PDFs referenced in the log (for example, guide pratique 54_expérience data.pdf or archimag-338v3.pdf), so we can try reproducing this internally.
3. Please try this once:
   • Close TheBrain
   • Delete the current Output.log
   • Reopen TheBrain
   • Start indexing the affected Brain again and let it run for as long as possible
   • Then send us the new Output.log
   • This will give us a cleaner log with the startup context included.
4. As a comparison test, please let us know:
   • Does indexing complete successfully on a smaller Brain?
   • Does indexing work on a brand-new test Brain?

That will help determine whether this behavior is specific to this one large Brain or more general.

Regarding CPU usage from the dekutron process: high CPU activity during indexing is expected, especially with many large attachments. We’ve also made internal adjustments to allow longer processing time before an index is marked as stuck, which should help in upcoming builds.

Thanks again for your patience and for helping us dig into this. We’ll keep you posted as we learn more.

Best regards,

John