SABnzbd Forums

I use Astraweb as my primary server and I have a block account setup as a backup. This used to work fine until a few weeks ago when the backup server just stopped kicking in on missing articles. I just switched my primary from Astra to a different server that a friend let me use his l/p for testing purposes. The backup server kicked in just fine with my friend's server as primary.

It is possible Astra changed how they report missing articles in a way that makes SAB not go to the backup? Or it is just me having this problem and it's a setting I might have forgot to check?

Astra has been known to use strange methods regarding DMCA takedowns.
But I'm not sure they still do.
I'm using Astra as one of my servers, but apart from some time-outs, it's not a nuisance.
Note that nowadays we use server priorities. 0 is highest.

We are not returning any unusual messages.
If an article has been removed, then we return something like "430 DMCA removed".
@MrLyle: If your problem is persistent/repeatable, message me and we (astraweb) will have a closer look.

I'm gonna keep an eye on this next time I hit a broken release and check my counters to see if my backup kicked in or not. I know about the priority system. I use 0 for Astra and 1 for my backup. This used to work perfectly for a good few weeks before suddenly just stopping. I'm pretty sure it's nothing from my end.

Did you do a server connection test in Config -> Servers for your backup server? Just to make sure all is well.

Yeah, the test passes without any problems. I had a release that was broken on Astra, files coming in with missing articles all over the place. The backup server did not kick in at all. I disabled Astra and used just the backup server to see what would happen. All the files came in complete. I stopped it after 5 or 6 files, didn't wanna burn my block account.

Seeing as how those files were complete on my backup server, it should have been used to complete the broken files from Astra. For some weird reason SAB isn't switching servers on the fly. This used to work perfectly for a good few weeks, I'm not sure what changed but I know it's not something I did. I haven't touched the configuration in a long time.

Do you have log level set to Debug by any chance (in the Status window)? If not, could you set it to Debug and download that job again, with Astra and backup enabled.
Then send me the log at [email protected]?
You use a recent (1.1.x) version of Sabnzbd, right? Then the logs will include an anonymous (without username password) version of your settings at the end of the log that we can use!

Ok, so I was using 1.1.0 and noticed 1.1.1 was released so I updated before retrying the job. I went to status and switched to +Debug and restarted the job. I let it run completely through and saved the log. Emailing it to you know. I hope I did it right. If I didn't, let me know and I'll redo it properly. Thanks for looking into it for me, it's much appreciated.

Edit: the email address was bad in your post. I corrected it to [email protected] which I assumed is what you meant. If it isn't, let me know and I'll resend.

Oops typos!
We seem to be having a problem with that address though.. Could you maybe send it also to [email protected]? Thanks!

Sent

Apologies for reviving an old thread but I’m recently experiencing the same problem as the OP and I note that there was no resolution posted to this issue.

In summary, just like the OP, Astraweb is my primary server and my backup server has recently stopped kicking in for missing articles.

Background: My ISP has a peering agreement with Astraweb. My primary server is news.<ispname>.net but the connections are directly to Astraweb and access is unmetered. I have a TweakNews backup account which is metered. For the 3 years I’ve used this combination, it has worked very well with around 3% of data coming from my TweakNews backup account.

I noticed recently that the TweakNews backup account wasn’t being hit at all. This change seemed to have happened around 3-4 September 2017 (thanks to the handy graphs in the latest version of SAB). Since that time, I’ve downloaded around 250 nzbs with 10 failing. These nzbs run the gamut from hours old to 900+ days. However, not a single byte has been downloaded from my TweakNews backup account, which is highly suspicious given the long term average of 3% of data coming from TweakNews.

I know the TweakNews server is working. As well as "Test Server" being successful, If I reverse the priority of the servers in SAB (ie set TweakNews to priority 0 and news.<ispname>.net to priority 1), data downloads from the TweakNews server.

As the OP reported, I suspect that Astra has changed how they report missing articles in a way that makes SAB not try the backup server. I'm not sure if this is a problem with Astra or SAB. As noted above, it started for me around 3-4 September 2017.

I’m currently running SAB v2.3.0 on a Synology NAS.

I’m keen to get this resolved and I’m willing to troubleshoot.

Thanks,
SABfaninAus

Alright, can you enable Debug logging in the Status window. Then pause your Queue (to avoid mixing up of logging from multiple jobs) and only download 1 NZB that was reported missing on Astra and that wasn't tried on Tweak.
Then after it fails get the log using Show Logging, and send it to me at [email protected]

Thanks for the extremely prompt reply! The log and failed nzb have been sent!

So the problem with this NZB specifically is that the articles are not missing, they are damaged.
In the log you will see CRC errors and in the download-report (click on the job in History) it will show "x malformed articles" instead of "x missing articles", meaning that the content is not equal to what it should be.
When these kind of errors are occurring, SABnzbd won't try the articles on a different server since it usually means that the upload didn't happen correctly or that the syncing between the uploader's server and the rest of usenet failed.
This is also the case for this NZB, I tried it on 3 different servers with different backbones (UNS, Frugal, NextGen) and all of them report the same number of malformed articles.
So maybe it's a coincidence that you are downloading from the same uploader and all those failed.

If you (or somebody else with this problem) do find a job where another server can complete the download without malformed articles, let me know and maybe we can implement something in SABnzbd.

Thanks for looking into this so promptly. Reviewing the 10 jobs which have failed since 3 September 2017, 9 of them report "malformed articles" and the 10th is most likely DCMA'd. Are malformed articles treated the same way as missing parts, in that PAR files can recover them? A lot of other nzbs I downloaded also had malformed articles and could be recovered.

While these malformed articles explain the failures, that still means around 240 nzbs (around 0.5 TB of data) have been successfully downloaded since 3 September 2017 without a single byte coming from my backup server, which is surprising to say the least. Maybe I've just been lucky?

Anyway, if I note a failed job with no malformed articles where my backup server is not hit, I will pass it on for investigation.

Thanks again.