FAILED jobs

[zaadstra]

Hi,

I've been searching here and puzzling the switches but I can't find what I'm looking for.

I am running on Synology nas and when upgrading the device I also upgraded from version 0.5.6 to 0.7.7.
The version 0.5.6 did save failed jobs (incomplete) with prefix FAILED_ and put this also in the "complete" folder.
Now version 0.7.7 seems to forget these jobs and list them in the history with a Retry option.

Did I not find the right configuration, or is the FAILED_ thing removed from v0.5.6 to v0.7.7 ?

[AutomaticCoding]

I'm still on 0.7.6 (Updating requires me to go to my box, I'm lazy. I'll get around sometime or another, but, the changelog isn't too big) and FAILED_ still seems to be in place, considering there's quite a few folders located in my movies "to organize" folder (for couchpotato) with the prefix "_FAILED_".

EDIT: Proof (Unless couch did this, which is a possibility, but, I doubt):-

[zaadstra]

Weird, I'm still experimenting but no FAILED_ prefix.

I've changed the post processing switch "Process only verified jobs" to unchecked, the jobs now are passed to the "complete" folder. They do not indicate "Failed". The notification mail also says completed, the mail says 77 blocks short, but also Enjoy! :-?

I must note, that SAB may have difficulties here with my NZB's, as some of them contain more than one sets of par's. Example: a set of mp3's including par2 and a set of rar's including par2. This is a .nzb fron Binsearch. Now, the mp3 par's are OK, and the rar par's fail.
SAB 0.5.6 reported the job as failed, and reported in the notification mail the mp3 set as complete and the rar set as failed. The job was put in the Complete folder with prefix FAILED_ to indicate the issue.
Now with current settings in 0.7.7 the job is passed to Complete, but not indicated as failed, and after pressing Retry in the History another set Rar's was downloaded. After messing around with my test job I ended with 5 of the same rar sets (that quickpar checks as OK) .... About 2GB downloaded for a job less than 500MB.

I was using the classic darkblue webinterface but changed to Plush to eliminate problems or miss extra options.... No change.

I also wonder why I see +Delete in the queue, I expected to see +R as I did not enable the unpack (and delete) options.

Edit: a created testjob with only one set of mp3's and one par2, with some missing tracks is reported in notification email header as completed. In History in All view it is reported Repair failed, missing blocks. and in View Failed it does not show up. Seems not right?

[shypike]

Since 0.6.0 jobs that fail to verify are kept in the "incomplete" folder.
The idea is that you retry the job with an NZB that contains more PAR2 files
(reusing the already downloaded files).
_FAILED folders are only made when the job did pass verification/repair successfully
but failed to unpack for some reason (like disk full).

[zaadstra]

Since 0.6.0 jobs that fail to verify are kept in the "incomplete" folder.
The idea is that you retry the job with an NZB that contains more PAR2 files
(reusing the already downloaded files).
_FAILED folders are only made when the job did pass verification/repair successfully
but failed to unpack for some reason (like disk full).

I see. That's a fundamental change between versions, I did not pick that up from version histories.

Isn't there a way to let these jobs go to the complete/finish folder with FAILED_, most time I know there won't be anymore Par files,
or it will be a new job later on. Now it takes time to manage this, while v0.5.6 just went on and signalled problems.

[shypike]

You can easily remove failed items, including their files.
You do this from History and there's even a scheduler event for it.

[zaadstra]

You can easily remove failed items, including their files.
You do this from History and there's even a scheduler event for it.

I don't want to delete them, as I explained above, Binsearch sometimes generates a NZB with multiple versions of the download.
Of these sometimes 3 or 4 present versions, usually most of the downloads do verify ok so I can use them In advance you never know what is
complete these times so I bet on multiple horses So the 0.5.6 system worked out better for me, but new fast Nas, new challenges!

[grambo]

I've been getting this quite a bit of late too. The files are successfully being extracted, as i see them in the completed folder, but they get tagged with FAILED_ and do not continue to run the post processing script. I'm running sab 0.7.7

[shypike]

Does it say why the thing failed?

[zaadstra]

Does it say why the thing failed?

Yes, not enough repair blocks.

An example: v0.7.7, multi-nzb (e-mails says SABnzbd has completed job ...)
Completed 2013-01-08 08:40:48
Name David_Guetta_-_Nothing_But_The_Beat_2.0-2012-MOD - Repair failed, not enough repair blocks (12 short)
Size 447 MB
Status Completed
Category *
Path /volume1/download/sabnzbd/complete/David_Guetta_-_Nothing_But_The_Beat_2.0-2012-MOD
Download Downloaded in 1 minute 1 second at an average of 7.3 MB/s
Source /volume1/download/sabnzbd/nzb_in/David_Guetta_-_Nothing_But_The_Beat_2.0-2012-MOD.nzb
Repair [David.Guetta.-.Nothing.But.The.Beat.2.0-2012-MOD] Verified in 53 seconds, all files correct
[David_Guetta_-_Nothing_But_The_Beat_2.0-2012-MOD.rar] Repaired in 1 second
[00_david_guetta_-_nothing_but_the_beat_2.0-2012] Repair failed, not enough repair blocks (12 short)
Trying SFV verification

v0.7.7, single-nzb: (e-mails says SABnzbd has completed job ...)
Completed 2013-01-07 22:58:21
Name Roxette-A_Collection_Of_Roxette_Hits_Their_20_Greatest_Songs-2012-C4-TEST-11-14 - Repair failed, not enough repair blocks (284 short)
Size 138 MB
Status Completed
Category *
Path /volume1/download/sabnzbd/complete/Roxette-A_Collection_Of_Roxette_Hits_Their_20_Greatest_Songs-2012-C4-TEST-11-14
Download Downloaded in 23 seconds at an average of 5.8 MB/s
Source /volume1/download/sabnzbd/nzb_in/Roxette-A_Collection_Of_Roxette_Hits_Their_20_Greatest_Songs-2012-C4-TEST-11-14.nzb
Repair [Roxette-A_Collection_Of_Roxette_Hits_Their_20_Greatest_Songs-2012-C4] Repair failed, not enough repair blocks (284 short)


Example mail of 0.5.6 with mulit-nzb, some compleet and some failed:
title:SABnzbd has failed job VA-Trance-The-Ultimate-Collection-2012-Vol-2-2CD-2012-wAx‏
Hi,

SABnzbd has failed to download "VA-Trance-The-Ultimate-Collection-2012-Vol-2-2CD-2012-wAx"

Finished at 2013-01-06 13:33:43
Downloaded 1.70 GB

Results of the job:

Stage Download
Downloaded in 21 minutes 44 seconds at an average of 1.33 MB/s

Stage Repair
[VA-Trance_The_Ultimate_Collection_2012_Vol_2-2CD-2012-wAx__www.realmom.info__] Verified in 49 seconds, all files correct
[000-va-trance_the_ultimate_collection_2012_vol_2-2cd-2012] Repair failed, not enough repair blocks (349 short)
[VA-Trance_The_Ultimate_Collection_2012_Vol_2-2CD-2012-wAx] Repair failed, not enough repair blocks (1896 short)

Sorry!
.... and folder moved to /complete with prefix _FAILED_

I understand the change in the handling of failed jobs. But leaving them in the history while I can't get extra parts any way takes much more post processing time for me (checking history, getting from /incomplete) than having it in the complete folder with an _Failed prefix indicator. Then I know the other folders are good anyway. Now I switched off postprocessing and all folders in /complete appear to be good. I wish I can swith on the prefix _Failed_ there.

Perhaps in the case of multi-nzb's above, the correct history line and notification text would be Partially good, partially failed. In case of single nzb as Roxette text would be Failed.

Also, I the postprocessing working correctly anyway? Yesterday I looked why I still see +Delete in the queue list, and when browsing the options I did not see why, and there is no option to switch on or off +Repair (it is repairing anyway). Looking in my 0.5.6 install there just are the clear +R +U and +D options.
When reading this it looks like newbie questions, but I've put some serious testing time in it


About the notification mails: I wonder if it is possible to add a status line with queue size (count and size) and disk free size?

[shypike]

Multi-part NZBs are a bit of a problem, which isn't going to be solved in the short term.
For each job you can still chose a post-processing option like you always could.
The behavior of 0.5.6 is not coming back, because the new one is better for most users.
You can still chose to disable "postprocess only verified jobs" inb Cfg->Switches, which more or less restores the old behavior.

[zaadstra]

You can still chose to disable "postprocess only verified jobs" inb Cfg->Switches, which more or less restores the old behavior.

Yes that's what I did

However I do miss the indicator that it is failed. I think it's remarkable good that all separate parts of such a nzb are processed and repaired if
needed. No change needed there!

Perhaps an idea to make a Failed flag option in a next version?

[shypike]

The idea behind disabling "only verified" is that you have a script that
handles this situation. One of the script parameters says what's wrong with the job.
So a simple rename script would do.