Bug discovered: Running .sh scripts after job completion causes inconsistent result message

Catfriend1 · May 6, 2021, 12:01pm

Hi,

I’m on Duplicati 2.0.6.1-beta. I’ve moved my post job .sh scripts to another location. When running “check files” on the job, it popped up a dialog box “1 warning occured”. Opening the job log, it said “green check mark - success, 0 warnings”.

The bug from my point of view is, that the reported warning/success are inconsistent (and reproducible) on the Web UI if the post-job run shell script xxx.sh does NOT exist.

I expect the following:

When the post job sh script does not exist, throw a warning and clearly tell the user that the problem is with the .sh file. The currently shown “1 warnings” with nothing in the log but “success” is not pointing the user at the root cause (I first misunderstood and thought there would have been strange problem with my job files which turned out thatwasn’t true.)
If the warning because of missing .sh post job script occurs, properly log it in the local job log and replace the “green ok check mark” with the “warning sign” accordingly.

Thank you.

Kind regards
Catfriend1

drwtsn32 · May 7, 2021, 1:17pm

Yeah, this is confusing. When an error occurs in some of the post backup tasks (script, email notification, etc), you may get a Warning message in Duplicati but overall the job shows as successful. I think this is because the main phases of the backup were fine (files read on the source disk, changes saved to the back end).

To see these post-backup error you can try watching the Live Log. It’s in About → Show Log → Live and set the dropdown to Warning. But you have to check that as the warnings happen or soon thereafter, otherwise the events get replaced by newer ones.

ts678 · May 7, 2021, 1:41pm

To answer better would require a logging expert, but there are none readily available. Maybe related:

Combine logs in GUI #1152

For technical reasons there are multiple logs.

This is really confusing to the user as they do not understand why and what happens underneath.

My guessing about this is that the backup makes a log from its results, and that’s the end of that log.
Clearly it has to be in some state of “done” to be emailed, but if email fails, it’s not getting updated…

I’d also note that the run-script-after script itself gets info it may want about short and detailed result:

github.com

duplicati/duplicati/blob/50ed059e11efeb73a5d0943f5349f64064e19c8c/Duplicati/Library/Modules/Builtin/run-script-example.bat#L116-L118


      
          REM DUPLICATI__PARSED_RESULT

          REM This is a value indicating how well the operation was performed.

          REM It can take the values: Unknown, Success, Warning, Error, Fatal.

github.com

duplicati/duplicati/blob/50ed059e11efeb73a5d0943f5349f64064e19c8c/Duplicati/Library/Modules/Builtin/run-script-example.bat#L101-L104


      
          REM DUPLICATI__RESULTFILE

          REM If invoked as --run-script-after this will contain the name of the 

          REM file where result data is placed. This value cannot be changed by 

          REM writing it back!

If a backup fails, there are no backup result statistics to send, so the next best spot is the server log.
Unfortunately the GUI sends them to the backup log, user sees nothing, and forum has to help them.
I would love to see this fixed, but I assume it’s a large job because the lead author has not in 7 years.

Catfriend1 · May 7, 2021, 1:49pm

Thanks for your good replies, very helpful to me :-)!