6.70 final still stalling

[RayMark]

Unfortunately, downloads are still stalling with cache = 0 in 6.70 final.
As far as I can tell, this is not the case with virus infected files. Just when downloading regular rar/par2 file sets.
Sometimes NewsBin stalls just after 1-2 hours. Even when cache is set to 2000.
Once that happens, even Help->About does not help - even though cache is increased back to 1500 or so, NewsBin stalls again after just minutes.

So, unattended downloading is not possible. Closing and restarting is needed several times a day.

If that problem cannot be solved, perhaps it is possible at least to detect it and to "restart" Newsbin automatically, preferably without closing open groups, i.e., an "internal restart" if possible. If not - a real full automatic restart. It would at least allow an unattended downloading - like it was possible with 6.55.
If somebody does not like it, it could be enabled/disabled in settings.

For full automatic restart, it is important to use the same specific path to the .exe file (because various different versions may be used) and the same .nbi file.
I often run at least two instances of NewsBin (sometimes different versions) with different .nbi files (pointing to different data locations, so there are no conflicts with headers or various other db3 files)
Those .nbi files are never the same as detected while installing Newsbin.
So again, restart with the same exe and with the same .nbi and do not close other/wrong NewsBin instance or all instances. But better to "restart" internally, if possible


OR PERHAPS:

Can you find out why once cache goes to 0 and downloads stall, further Help->About have only very temporary effect? Perhaps you could fix that.
Then, catch all suspicious exceptions and clear the cache. Perhaps an option in settings - to clear the cache on all exceptions. But that would work only if clearing cache really works.
Currently, it seems that there is another issue that goes together with cache = 0. Help->About only clears the cache but does not reset something else that has to be reset or re-allocated.

[Quade]

I can't remember if you tried it on a different PC or not.

I'm unlikely to build any kind of automatic restart for something like this.

How many servers are you using? Does it happen if you only use one server? I'm wondering if you have a dead server in the mix and all the bad chunks are waiting for this server that never downloads anything.

[RayMark]

I am using 3-4 servers. One server does not renew subscription automatically so I just disable it then.

Yes, I noticed such occurrences with a dead server. BTW, it is probably sort-of-a-bug in Newsbin if a dead server blocks all downloads - the item in the list requiring that server could be (temporarily ?) skipped . As it is now, it is bad for unattended downloading.

But I don't think this is the case, though. In the case of a dead server, does Help->About help even if for a short while?
But I should double check. There was one especially bad day with maybe 10 manual Newsbin restarts - that perhaps could suggest a problem with a server that day.
So I will check the next time.

I am somewhat prone to situations with a dead server. I am using VPN for torrents, and when the VPN connection is lost and VPN automatically reconnects, the IP address can change and servers do not like that.
I do set the fastest servers through a different ISP directly, but not all of them, and sometimes they jump to the VPN anyway. Another issue is multiple NewsBin instances, If the number of simultaneous threads in all the instances exceeds the maximum allowed number for a server, then even when playing with "pause", "too many connections" do happen...

[Quade]

es, I noticed such occurrences with a dead server. BTW, it is probably sort-of-a-bug in Newsbin if a dead server blocks all downloads - the item in the list requiring that server could be (temporarily ?) skipped . As it is now, it is bad for unattended downloading.

Some people pay "per byte" to use some servers. They wouldn't take kindly to me running through all their bandwidth because of a connection problem. If you have a dead server, it's up to you to manage it.

I am somewhat prone to situations with a dead server. I am using VPN for torrents, and when the VPN connection is lost and VPN automatically reconnects, the IP address can change and servers do not like that.
I do set the fastest servers through a different ISP directly, but not all of them, and sometimes they jump to the VPN anyway. Another issue is multiple NewsBin instances,

I'd turn all this shit off, download with a single good server and see if it still stalls. Then add a second server and try again.

If the number of simultaneous threads in all the instances exceeds the maximum allowed number for a server, then even when playing with "pause", "too many connections" do happen...

Yeah, don't do that.

[RayMark]

Does not look like a dead server.
The error messages seem to be the usual ones, except those two:

1. Error talking to Server - it does not look important, because only very few of those, not continuously, but perhaps (sometimes?) it can cause a chain reaction - internal errors in NewsBin?
2. - just an empty red line - ? - could it indicate an internal error in NewsBin? Especially that it ALWAYS follows "Error talking to Server" error - but shows up not always at once after that error, sometimes after some time, it seems.

But only a few of those, so not continuously blocking downloading. And cache = 0.

Interesting, why once this happens, Help->About does not fix it completely, but just temporarily - the cache afterwards goes continuously down until again downloads stall at cache = 0.
This happens much faster than the first time after restarting Newsbin.

After it happens again, there are only a few lines in the log, again, nothing blocking downloading continuously.

[Quade]

Did you do what I suggested? Single server? No VPN?

[RayMark]

Not yet.
I cannot afford to stay without VPN for more than a few hours, so no stalling might mean nothing.
On the other hand, routing nntp not through vpn usually works fine. I can see clearly through which ethernet port (ISP) the traffic comes, besides, if the download speed is very high, it is unlikely to be through VPN anyway.
However, sometimes nntp jumps over to vpn. Probably after the other connection was temporarily down.
So, I guess I could leave one server and make sure that it is not through VPN.

Do you think the problem might be with multiple servers even when none of them is dead?

[Quade]

Do you think the problem might be with multiple servers even when none of them is dead?

No clue at this point. You have too much going on to even guess. The only way you can trouble-shoot this is to isolate the download and test each step till is starts to fail.

[RayMark]

Is it possible to disable the cache completely?

I tried to increase its size from 200 to 2000. Is it possible to set its size to 0 so that it is not used at all? Or some other way to disable its usage?
If so, perhaps the issue of the stalling downloads would go away.
Also, I feel as if using the cache slows down downloads almost two times: first, rar parts are downloaded. But they do not appear as files. Some other previously downloaded post is gradually appearing instead.
Its like playing with a time-shift enabled. Finally, when they begin to appear, they appear slowly, one at a time, almost as slowly as the initial download. So it takes twice as long for rar parts to appear compared to the old good version 6.55.
Also, another annoying problem - when I need something to be downloaded immediately and move it to the top of the list, it is downloaded but does not appear as file. Many previously downloaded items keep appearing instead.
So once things are "downloaded" (but probably only to the cache, so to me - not yet downloaded) other items cannot be moved in front of them. I need something quickly, but I have to wait for hours instead.
So, if the cache is disabled, then this problem would go away as well.
The only remaining problem would be "pending" in the download list - some pending items are not resolved quickly enough, I need to wait and wait and wait. I am afraid that disabling the cache would not help with this.

To solve all the problems at once, I need to use 6.55. It is the best version by far. However, 6.55, although almost perfect, it still has one issue, especially when downloading books - incomplete/damaged files are not correctly marked with all capitals but appear as regular correct downloaded files.

[Quade]

Also, I feel as if using the cache slows down downloads almost two times: first, rar parts are downloaded. But they do not appear as files. Some other previously downloaded post is gradually appearing instead.

It's not slowing anything down. The download is so fast, it's faster than the combination of unrar and assembly. If you want to keep it from running ahead, enable "Pause download while unrar/repair" in the performance options.

How about trying Newsbin on a different PC without all the extraneous junk running on it?

[Quade]

I have a version I want you to try. Let me run it by Dex first.

Be pretty soon.

[RayMark]

OK. is it 6.71 beta 2?
It does not look like it from the list of changes

The other issue - not being able to download something quickly - I think I should post it separately as a bug.
Even if I use "Pause Download UnRAR/Repair", downloading still runs far far ahead of extracting.
If I need something to download quickly, putting it at the top of the list does not help. It is downloaded, but it is not decoded and extracted. Instead, other previously already downloaded files are decoded and extracted.
Even if I put them in "pause" mode - it makes no difference: they are still extracted before the file at the top of the list that I desperately need as soon as possible.
So I have to wait for hours for a file that I need at once. Because I sometimes put into the download list quite a lot for unattended downloading.

Currently, the only solution: to start a different NewsBin Instance.
You really should make an effort to move also already downloaded stuff according to the order in the download list.
Or at least, make sure that nothing further is being downloaded until the already downloaded thing is decoded (moved from cache to the download folder), repaired and extracted.
Do not download ahead, if it is not possible to make the decoding/extracting order to reflect the order in the download list.
Still, a failing download/repair/extraction should not halt all the further downloads forever. Problematic stuff should be skipped

[Quade]

I think I forgot to PM a link.

http://www.newsbin.com/downloads/nb671B3.exe


I'm ignoring everything else except stalling. Until that's solved, there's no point discussing anything else.

[RayMark]

6.71 beta 3 does not start. It remains running as a background process but never shows up on the screen.

I have installed it and tried to run it while another NewsBin 6.70 final was still running, though. It still is.
Maybe I am cutting corners too much.
Besides, I probably should try with various db3 files removed (Downloaded.db3, Downloads.db3, etc.)
But there is no problem with starting the second 6.70 final instance.