Problem on CentOS Stream 9 - OCRmyPDF succeeded with warning(s): sh: line 1: ocrmypdf: command not found,

[mrojrti]

When I try to install ocrmypdf via dnf/yum the packet was not found on the repository even epel installed.

Tried via Snap and Pip but I get the same error above on Nextlocu log:

Can you help me, please?

Thank you

[R0Wi]

Hi @mrojrti, I can't immediately tell you what's wrong but I'd suggest you try the following steps to get closer to a solution:

1. Login to your server with your regular user (or root) and check if ocrmypdf --help produces any reasonable output. If not, it seems like the installation is somehow completely broken on your system.
2. Since OCR jobs are executed asynchronously via NC cron: check the way your Nextcloud cron is executed. If it's triggered via system cron, check which user executes the cron.php script. If it's executed via Webcron or Ajax, the user will be the same one running your webserver (in that case, check who is executing apache2 / httpd). If you identified the system user executing your NC cron, open a new shell for that user and try step 1 again (for example sudo -u www-data bash and then again ocrmypdf --help). If that works, everything should be fine. If not, the executing user of the NC cron doesn't have access to the ocrmypdf binary. Try to make it accessible globally on your system by putting the directory where the ocrmypdf binary resides into your PATH or try something else so that this user can properly execute the ocrmypdf command

Hope this helps!

[mrojrti]

Hi @R0Wi !

All right, I redo some tests and the user apache on terminal can't run the commando ocrmypdf, returns command not found.
When I tried the cron with root, I get this message on log:

So I do: crontab -u apache -e and add: */1 * * * * php -f /var/www/html/cron.php
Now no errors on cron log:

When I run the sudo -u apache ocrmypdf the command returns ok (I change a security configuration on sudoers to get this work)

But still not working on Nextcloud, command not found =(

I managed to get the command work on apacher user, but I don't know how to proceed now, can you help?
Any idea?

Thank you so much.

[R0Wi]

That really sounds like a taff one 😄

To get closer to a solution I'd not work with the crontab but instead call cron.php directly, think it's better for debugging. So:

1. Make sure your NC cron is configured to cron
2. Make sure the apache crontab is the only one with something like php -f /var/www/html/cron.php, so that really nothing else executes the cron.php
3. Remove or comment out the crontab entry for apache for the time debugging
4. Upload a new file which matches your OCR flow
5. Open a new console and change your user to apache (sudo -u apache bash) then execute php -f /var/www/html/cron.php.
6. Inspect the logs
7. Don't forget to uncomment the crontab after testing

It would also be interesting to know where exactly the ocrmypdf binary resides on your system, because it has to be somewhere, where not every user has access to or doesn't have this location in his PATH. With a user where the ocrmypdf command succeeds, execute whereis ocrmypdf. On my system I get something like:

docker@489f65697e04:/var/www/html$ whereis ocrmypdf
ocrmypdf: /usr/bin/ocrmypdf

So you see it's below /usr/bin, which is part of the system PATH variable. If that's not the case on your system, another approach would be to create a symlink under /usr/bin which points to ocrmypdf (something like ln -s <your-path> /usr/bin/ocrmypdf). This should make the command globally available.

Good luck!

[mrojrti]

Hi, @R0Wi
I hope you're well.

Finally, it works!!!

1 - The cron config was ok.
2 - Yes, it was the only one
3 to 7 - On this test the problem continue.

Executing whereis ocrmypdf the result was the expected and I had the symbolic link to.

Resolution:
I put the crontab on the root user, but with sudo -u to apache, like this
*/5 * * * * sudo -u apache php -f /var/www/html/cron.php

Now it's working like a charm.

Thank you so much for your attention!

[R0Wi]

Happy to hear things are working now. Sounds to me indeed like in some situations the apache users PATH isn't setup correctly. Good to know 👍

[mrojrti]

Yes, I forgot 1 step that I did
As root I ran visudo and add the path /usr/local/bin of ocrmypdf to secure_path.

Thank you!