0.4.9 Lockups when one server is down.

[SenorClean]

I've noticed in 0.4.9, 0.4.8, 0.4.7 that if I am using multiple news servers, and one is down - SAB will download for a little while with the working server, then lock up totally.  The speed reported is constant, and no downloads progress.  If the process is ended, and restarted - it will download for a little while longer then freeze again.

Sample warning from a server I receive when it is what I am calling 'down' -
2009-04-21 14:26:29,220 ERROR [newswrapper] Failed to connect: (10061, 'Connection refused') [email protected]:119

Removing all servers, re-adding the one(s) that are up only, and restarting SAB fixes the problem.

I'm running 0.4.9 under Windows 2008 - and have noticed this behavior also in Vista.

[shypike]

This will not improve until we release 0.5.0.

[SenorClean]

Thanks for your reply - is this a known issue?  Does it only occur in Vista / 2k8?

Let me know if you would like logs - or if there is anything I can do to help.

[shypike]

It's a design problem.
Especially the combination of a non-reponding server and a missing article creates problems.
It requires a redesign which we are not going to do for the 04.x series anymore.

[pip]

I've noticed in 0.4.9, 0.4.8, 0.4.7 that if I am using multiple news servers, and one is down - SAB will download for a little while with the working server, then lock up totally.  The speed reported is constant, and no downloads progress.  If the process is ended, and restarted - it will download for a little while longer then freeze again.

Sample warning from a server I receive when it is what I am calling 'down' -
2009-04-21 14:26:29,220 ERROR [newswrapper] Failed to connect: (10061, 'Connection refused') [email protected]:119

Removing all servers, re-adding the one(s) that are up only, and restarting SAB fixes the problem.

I'm running 0.4.9 under Windows 2008 - and have noticed this behavior also in Vista.

iv also discovered this bug, after you but before i found this MB section/thread

iv discovered that you can also set the troubled server as  a "Backup server:" (rather than remove it totally) and it will also continue as before, you do need to kill and restart the sabnzbd demon though after you have saved the config, and you can also crash/lock the demon/server if you try and change settings before pausing the currently active Que, so make sure you do that first....

shypike regarding it not being fixed in 4x , i wonder if the demon is locking before its reporting "ERROR [newswrapper] Failed to connect: (10061, 'Connection refused') [email protected]:119 "

if not, then couldn't this internal error report be auto redirected to some new code so as to pause the active Que ,forcefully stop the currently bad news server connection and set it to "Backup server:" and restart the Que as a temp fix ? , at least until 5* is working good enough for a windows compile!

this "bonding" of several news servers at once to pull more download bandwidth than a single server allows  is a very unique option for NNTP only found in SAB it seems, and something id like to take advantage of long term ( i am using it with IPv6 now), for instance setting up private mini news servers that end users can install and use , advertise their small collections pulled from the bigger servers and peering with other end users using their limited upload for community sharing of their various ISP WAN connections,and collective  HD space, old school Usenet run by the users as it were ,(non of this payed for commercial services that's got everyone thinking you cant do it yourself anymore) with SAB at its core , an internal small news/peering server like the old free "hamster classic" win32 .pas code base might be a very good thing to include in a later SAB ... but that's getting OT so ill leave that here.