OCR Timeout

[Alexires]

Hi all,

Would it be possible to build in a timeout for OCR processes? I find with some PDF's the OCR just hangs on a page while other PDF's are fine. If a timeout was built in, this would prevent server being under load waiting for OCR to finish (which it doesn't).

Ubuntu 10.10, OpenKM 5.1.9

[jllort]

I think we have added it in trunk pavila can confirm it.

[pavila]

This is implemented in the 5.1 branch, try 5.1.10 night build to test it. The timeout is hardcoded to 5 minutes.

[Alexires]

I think I fixed the problem by upgrading Cuneiform from 0.7.0 to 1.1.0 in Ubuntu via Aptitude. Good to know anyway. Thank you.

[pavila]

Depending on Cuneiform version and operating system, the program may fail. Cuneiform 1.0.0 and more recent version works fine, almosty in Ubuntu and Debian.

[Alexires]

Did this end up being included in 5.1.10? Can I change that timeout time?

[jllort]

As said pavila "The timeout is hardcoded to 5 minutes." can not be changed now from administration configuration I will add in our features ticket system http://issues.openkm.com/view.php?id=2215

[Alexires]

Having this in the admin configuration would be massively useful. As an example, for my system, a PID of the cuneiform process rarely runs for more than 10 seconds, so a timeout of 5 minutes is far too long (for my system). If I'm uploading many files, I need to sit there and watch htop and kill cuneiform processes that hang.

[jllort]

The temporary solution for you could be disabling ocr, and enabling before ... that will not index image files, but repository can be reindexed before.