NDMP job won't stop even when completely removed from catalog

[Andrea Bonetti]

Hi, 

i was experiencing this behaviour in version 23 and it's still there in version 24, every time a NDMP job gets stuck for some reason(usually no appendable volumes) and i cancel it bareos proceeds to inform me that the job is being cancelled, if i check the status of the storage i can't find it, but on webui it says "running" and if i try to start another job on that storage it waits because the NDMP one is still "running".
I even tried out of desperation to completely remove the job from the catalog, i couldn't see it in webui but jobs were waiting for it anyway(if duplicate jobs are not allowed starting the same job cancels it right away because the former one is still "running").
Docs says that NDMP jobs taking a while to stop is expected behaviour, but i found myself waiting 72+ Hrs for it to really stop and only a director restart solved the issue.

i'd prefer not to reload the daemon each time something goes wrong on the several NDMP jobs we have.

thanks

[Bruno Friedmann (bruno-at-bareos)]

Hi Andrea,

If you can reproduce the case, it might be interesting to open an issue on https://github.com/bareos/bareos/issues/new/choose

Please help any of us by documenting as much as possible the case with configuration, log extract, command and action used.

More the case is documented and explained, more the chance to reproduce it are bigger, more there's a hope to create a fix.

Otherwise, maybe you get lucky and it will be fixed in 25 ;-)

[Andrea Bonetti]

did it, thank you