3.4.2 Constant "Lost connection to SABnzbd" messages during downloads

[wcanyon]

I am getting the "Lost connection to SABnzbd" constantly when I'm performing simple tasks like downloading or deleting things that haven't started downloading yet. CPU load is low, under 15%. It's having a hard time loading the list of downloads as well. "Lost connection" has shown up 6 times since I started typing this. Downloads are only running about 1.5MB/s so I don't think it's swamped with file ops.

Rebooted the system about an hour ago, still seeing the issue.

What next for solving this?

Mint 20.2

Also a lot of this in the error window:

Code: Select all

File "/usr/lib/python3/dist-packages/cherrypy/_cpwsgi.py", line 189, in trap
    return func(*args, **kwargs)
  File "/usr/lib/python3/dist-packages/cherrypy/_cpwsgi.py", line 105, in __call__
    return self.nextapp(environ, start_response)
  File "/usr/lib/python3/dist-packages/cherrypy/_cpwsgi.py", line 438, in tail
    return self.response_class(environ, start_response, self.cpapp)
  File "/usr/lib/python3/dist-packages/cherrypy/_cpwsgi.py", line 244, in __init__
    self.run()
  File "/usr/lib/python3/dist-packages/cherrypy/_cpwsgi.py", line 347, in run
    request.run(meth, path, qs, rproto, headers, rfile)
  File "/usr/lib/python3/dist-packages/cherrypy/_cprequest.py", line 617, in run
    raise cherrypy.TimeoutError()
cherrypy._cperror.TimeoutError

I can't make much sense of that stack trace, maybe a middleware error? SAB is running on a local network in HTTP mode.

Also I'm deleting about 15 jobs while downloads are paused and it's taking about 10s per deletion.

Logs here: https://paste.kodi.tv/afidevivak.kodi

[sander]

While SAB is not doing anything:

In SABnzbd's upper right corner, click on the wrench symbol ("Status and interface options"), then click on first tab Status, and therev click on the Refresh Arrow. ... post the results in the lower part (no need for the IP addresses)

You seem to use a NAS (/mnt/nas/processing/incomplete/), so I wonder about that speed.

[wcanyon]

Code: Select all

Used cache  13.2 MB (18 articles)
System performance (Pystone)  98687  Intel(R) Celeron(R) 3205U…
Download folder speed  5.2 MB/s  (/mnt/nas/processing/inco…)
Complete folder speed  4.4 MB/s  (/mnt/nas/processing/comp…)
Internet Bandwidth  13.6 MB/s  (108.8 Mbps)

Last night when I tried to run those I got about 20 "Lost connection" messages and the values never refreshed.

[sander]

you get " lost connection" while SAB is not doing anything at all?!

[wcanyon]

https://imgur.com/BrIb7GC

That's a screenshot of the settings screen when the app is paused. You can just barely make out the "Paused" in the upper left corner. There were about 8 jobs in the queue. No post processing happening. Some parts of that image are blurred intentionally.

[wcanyon]

I think what I should do is move the incomplete folder on to the same drive and see if that fixes it. Still, network latency for a NAS mount shouldn't be causing a socket issue if you ask me. Plus, I watch movies over that network connection all the time and it's fine. No movies were being watched when these tests were happening.

[sander]

I think what I should do is move the incomplete folder on to the same drive and see if that fixes it.

Yes, please do so, to an internal, higher-speed drive: at least 100MB/s

Still, network latency for a NAS mount shouldn't be causing a socket issue if you ask me.

I agree, but let's check the result of using an internal drive. As they say: "meten is weten".

Oh, also, while SAB is doing nothing: what are the small numbers in the upper right corner saying? So: uptime CPU / queue values?

EDIT

PS: in your screenshot, I see download-spikes. So ... you were downloading while doing the test ... ?

[wcanyon]

No, no downloading while doing the test. It says 0B/s in the screenshot. Did you want a download test?

Here are the numbers after I changed to a local folder:

Code: Select all

Used cache  0 B (0 articles)
System performance (Pystone)  100503  Intel(R) Celeron(R) 3205U…
Download folder speed  70.1 MB/s  (/home/media/tmp)
Complete folder speed  3.5 MB/s  (/mnt/nas/processing/comp…)
Internet Bandwidth  8.9 MB/s  (71.2 Mbps)

The problem with using that ~/tmp folder is that the drive doesn't have much space on it. I can't download anything over 5GB or so on it. Sure, the drive is too small. I don't particularly want to have to move the OS over to a new drive just to be able to use SAB without these lost connection messages. I don't mind if my downloads take longer, they're all automated and run overnight most of the time anyway.

[sander]

I understant that. But we're trying to find a cause.

So, with Temp/Incomplete on your /home/media, while NOT dowloading anything, do you still get Lost Connection?

And, out of curiosity: as it's only 70.1 MB/s ... what kind of drive is that? Not an SSD, I guess?

[wcanyon]

Ah so it turns out that even with the incomplete folder being on the same drive I'm still getting lost connection.

I think I see the issue. The computer I'm accessing it from is seeing some lost packets when I ping the server that has sab running on it. So the client side is seeing a broken web socket connection but it's very brief. Seems like sab has a 10+ second timeout on retrying that socket connection, so it seems like a much bigger issue than it is. It's a new house and I don't have any access points wired up yet, so it's just the one router which is about 35 feet from this PC.

Yes, it's a very old SSD. I would consider upgrading to a newer one but that means reinstalling kodi, radarr, sab and my IR receiver (which is involved because the "current" repo for lirc has an invalid binary, you have to use an older repo) and getting my mounts back up and all that. 2-3 hours of work. It's working as is, so I'm not all that inclined to change it.

edit: actually that lost connection timer appears to be more like 30s.

[sander]

So ... packet loss on your LAN? Hmmm, that is not a great basis. Maybe solve that first? Or: in SAB: Wrench -> Webinterface -> Refresh rate to 1 minute or Disabled.

[wcanyon]

I think that socket connection timeout is not governed by that refresh rate. I am 100% sure that my connection the LAN is not dropping off for 30s at a time. It's one dropped packet in 8 when I ping continuously. That's not gonna explain the "Lost connection" being on screen for 15+ seconds.