SAB not working all of a sudden?

[rahduke]

I'm running ubuntu 10.10, have been using sab successfully for over 2 years now and everything has truly been great. For some weird reason SAB stopped working the other day, every time I add something to my queue it just idles until it times out. I checked my server settings and have viewed the logs but I just get timeout errors every time. I setup Lottanzb just to test my server and its working perfectly. I really am baffled as to what went wrong, I'm now on Sab 0.5.4 not sure if it recently upgraded itself from the repo's (new version maybe the problem), Any help would be much appreciated.

If this helps, its the last couple errors from my log files

"2011-03-24 18:55:05,593::INFO::[newswrapper:215] Failed to connect: (110, 'timed out') [email protected]:119
2011-03-24 18:55:05,594::INFO::[downloader:721] Thread [email protected]:119: timed out"

[jcfp]

Versions of sabnzbd prior 0.6.0b4 don't handle server timeouts very well, which can cause a complete stop of downloading. There haven't been any recent updates to 0.5.x packages. You could try the beta, available here.

[daleus]

Hi JC.

Could this be related to my issue?

(Debian 6.1 / SAB 0.5.4)

Connections (SSL - Giganews) take ages to spawn.

I have 50 connections and they can take 3-4 minutes to all connect up then downloads begin and go at full speed, until they either time out or finish sending the data.

At this point I then have to sit and wait for another 3-4 minutes for them to all connect again (or refresh or whatever they do) and another 700MB or so will download.

I tried turning SSL off and just using port 119 but it does the same thing. If I set connections to 10 they connect much quicker (more like a minute) and then I begin downloading but at a reduced rate (due to only using 10 connections).

This began when I formatted my server to Debian 6  and installed SABnzbd from the repo as usual - did not seem to be an issue on Debian 5.

I feel to point out that despite all the faffing once downloads are complete the files are fine, just to be sure I formatted server again and pointed the download location to an empty disk (to rule out my raid array as being dodgy).

Thanks for any help yourself or anyone else can give.


*edit, when I run sab I have the incomplete dir and complete dir on the same array, and during testing they were on the same drive - not sure if this would be an issue as it has never been before).

[shypike]

Using 50 connections is usually counterproductive.
Scale back until your speed just starts dropping.

[daleus]

Thanks for the reply, but that wasn't my issue - my issue is that each connection (on the connection page) takes ages to actually DO the connect. This means once I add an nzb the download will not start for about 5 minutes (on a 100Mbit connection this means it takes me 2.5 times MORE time to START the download than actually download the item. This never used to be the case at all.

*edit - you'll notice when I use 10 connections those 10 connections also take an awfully long amount of time to connect, but then download just fine, exactly the same as 50.


*double edit* - fixed. Adjusted timeout to 240 rather than 120, disabled 'only get articles from top of queue'. Now when adding an nzb it takes merely 4 or 5 seconds to start.

[shypike]

Like jcpf said, there's an issue with the 0.5.x series,
but that has mostly to do with unresponsive servers, which Giganews generally isn't.
So the timeout change, doesn't make much sense to me.
The "top of queue" off is beneficial in other ways (except for memory usage).
But if they help, that's cool.