newsbin.com

Lots of bug fixes and a new feature for Beta 5:

• Fixed "Use Recycle Bin" setting so it is saved between runs
• Fixed the Date/Age toggle (ctrl-t) so it is saved between runs
• Fixed an issue with the new duplicate detection database doing a conversion during startup.
• Added a new path variable called $(SEARCH) so you can use your search string as part of the download path.

The download link and more details about this release are available on the Newsbin Beta Page.

Please direct any questions or bug reports as a reply to this thread or through our Technical Support Form.

OK still hangs on startup. No notification of any mods going on. (Db conversion etc)
Hanging with only 90% of the screen painted (under wine so probably different under M$)

Can minimize it, but it isn't responding to clicks.

This is coming from b3 as b4 totally failed for me.
I'll leave it running for the next couple of hours as I am convering some mkv's anyway.

Conversion happens in a background thread now so, it's most likely some linux thing. You could delete the signature.db3. That would confirm it. Might be worth installing from scratch or installing to a new wine folder and see if it still does it.

Having tried the other four betas without success, I tried #5 and still get the same problem.

The problem is downloading from one particular group (a.b.e*****.) I click on the icon to download a file and it goes to the download list but after a couple of MB it stops. This time I've tried a few things to get it working and have succeeded but it is weird.

My setup is 2 usenet suppliers using 3 servers.

Main download server - Virginmedia.
Fill server - Easynews but using the US and Eu servers.

I setup a download and it stopped after a couple of MB so I unchecked the "Fill server" box on the 2 Easynews servers and the download started again from the Easynews servers and completed. The file in the download list disappeared but left a dotted outline across the pane with a red exclamation icon - the download was in the proper folder.

I checked the "Fill server" boxes again and setup another download and it stopped after a couple of MB. This time I unchecked the "Enable Server" boxes for the Easynews servers and the download started again from the Virginmedia server without any prompting. While the download was going on I re-checked the " Enable Server" boxes for the Easynews servers but the download didn't stop and carried on from the Virginmedia server. When the download completed the file name disappeared but again left a dotted outline across the pane with a red exclamation icon.

I then put 2 files in the download list and the download stopped after a couple of MB - unchecked the "Enable Server" boxes and the download started - re-checked them and the download carried on. The first file completed and the second file downloaded without stopping. Both files disappeared from the list with no dotted outlines.

The only other info I can give you is that when I tried the first Beta I noticed that the Xfeatures for Virginmedia was unchecked and I had to put the check mark in manually - for all the other previous versions this was done automatically as Virginmedia doesn't compress their headers/downloads.

About the dotted outline around the two completed download files - the header download for a.b.boneless was still going on while these 2 files were downloading. When I put the 2 files at once in the download list, boneless had finished downloading and there was no dotted outline and icon left when the files completed - I don't know if this has any significance but it's all info.

Installed Beta 5 over Beta 4, and Beta 5 crashes on startup every time I've tried to run it (4 times, then went back to beta 4).

I even deleted the .db3 files in the data folder and restarted. Still crashed on startup.

Beta 4 runs fine. Using WinXP SP3.

This was a new wine I created last time. Still hung now ~3hours later.
Will try again tomorrow but my sig db was only 7.5Mb anyway.

mimauk

Just for grins, disable the VM server, set one of the easy servers to fill and the other normal and try some downloads. Just a guess but, it sounds like the VM server is stalling but, because a stall isn't an error, the fill servers don't kick in. Fills only kick in when and actual error happens. No errors = no start.

As for the crash. I don't know. I'll have to find an XP machine to try it on.

Kenr, don't fixate on the signature.db3. I'm not convinced that's your problem. Easy enough to figure out, rename or move it and if it still hangs like this, the signature DB has nothing to do with it.

Quade wrote:mimauk

Just for grins, disable the VM server, set one of the easy servers to fill and the other normal and try some downloads. Just a guess but, it sounds like the VM server is stalling but, because a stall isn't an error, the fill servers don't kick in. Fills only kick in when and actual error happens. No errors = no start.

.

Ok I disabled the VM server set the Eu server as "main" and the US server as "fill" and set a seperate colour for each server. Put 2 files in the Download list and expanded them to see where they downloaded from.
The first file started downloading from the "main" server stopped after a couple of MB and then continued downloading from the "main" server.

I reset everything back to normal and put 24 files in the download list and they started downloading from the VM server but the speed was up and down from 100MB to 20MB for files 24,23,22,21 but it stalled again on file 20. Disabled the 2 Easynews servers and the download restarted with the speed up and down. Enabled the Easynews servers again and the downloads continued with the speed very erratic.

I use Bitmeter2 up in the corner of my screen to monitor my download totals and speed and the speed graph is up and down.

I added 2 more files from a tv group and the downloads continued with the speed erratic. For grins I unchecked the "fill" boxes on the Easynews servers and the speed smoothed out at 100MB - checked them again and the speed became erratic again.

The download stalled again with 5 files to go - disabled only one of the Easynews servers and the download started from the VM server and then stopped again. Disabled both Easynews servers and the download started with a smooth 100MB.

So definitely a problem with the decision making of where to download from as far as I can see.

Had 5 items in que of beta 4, all downloaded as normal. Que is now empty. Quit B4, installed B5 and the que now showed the previous items that had downloaded in B4, NB then started to download them as normal. I cleared que, quit and all seams normal now

OK done somemore testing this morning.
Changing/removing the sig.db did nothing, still hung.
Under wine settings if I disable "let the WM control/decorate the app" then it will load ONCE.
The second run it hangs same as above, no response, cant quit it or use force quit to exit, have to kill it manully.
Back to b3 for now, it works with without changing the above settings.

Have you upgraded wine or something? All the betas use pretty much the same GUI. The only change I can think of is that the menus change from tab to tab.

Maybe that's what it is, the menu's. Not sure why though.

Nope the wine is the same as for all the beta's hasn't been updated for a few weeks.
Tried it under crossover installed but didn't appear as an app for some reason.
Will try in a completely different wine folder but I have already done that for b3 to work.

OK completly new wine folder.
Installed b5, moved over my data folder, launched b5 it launched, then I registered it and quit.
Re-ran it and it hung, no responce to mouse at all, quit failed had to force quit it.
Don't really want to have to reconfigure everything to test this, but will move back the data folder and see if it launches repeatadly.
EDIT:
Moved the data folder back and launched b5, as expected it came up with the default layout but hung again. So not something in my data folder that is causing the hang.
Will try removning the WM bits in wine again.
Tried that and it hung completly coundnt force quit it, had to use kill pocess to remove it.

Ok I disabled the VM server set the Eu server as "main" and the US server as "fill" and set a seperate colour for each server. Put 2 files in the Download list and expanded them to see where they downloaded from.
The first file started downloading from the "main" server stopped after a couple of MB and then continued downloading from the "main" server.

So, does this mean it worked as long as you don't use the VM server?

I wanted you to run some real tests just using the Easynews servers. If it works with the easynews servers but, not when the VM server is in the mix, I'd suspect some weirdness with the VM server. That's what this is sounding like. The fact you get funky speeds when downloading solely from the VM server also suggests something is up with the server. As I said, stalling won't trigger a fill server to kick in.

Are you paying per byte on the easynews servers? Just wondering why you don't just leave them enabled and not fill.

I only got the funky speeds when both VM and Easynews servers where in the mix - isolate either one and the speed smoothed out.

I'm only guessing how NB works this but it seems to me that a when a file starts to download, the settings call for it to download from VM as it is the main server so it sends a request to VM for the first tranche of the file and when that appears NB wants the second tranche but then gets stalled by not knowing which server to send the request to. If there is only one server in the frame there is no problem but with two NB becomes indecisive.

I see there is another post about multiple servers from someone else in the UK I don't know if this has any bearing on it.

I just pay $10/month for 20GB on Easynews as although VM usenet is free with my deal it only has 7 days retention.

As a long shot - on the server list, the VM server is at the bottom of the list even though it is the main server - would that make any difference?? or is the list just alphabetical?

Order makes no difference.

it sends a request to VM for the first tranche

You sound like a financial dude.

Each chunk is a discrete element. When a server asks for some work to do, non-fill servers just grab the first available chunks and processes it. In the case of a fill server, the chunk isn't touched until it has one confirmed fail. Meaning say, Newsbin requests 10 block from VM and it's just sitting there waiting for the data from the server, the fills will do nothing until the data is returned or the connection times out. You can see this in the block display. If N blocks are showing 1/2 color and just sitting there at 1/2 color, it means the threads are waiting for data from the server. It's not a fail at that point so, the fill will just sit there doing nothing.

After 4 minutes or so, the connections to VM should start timing out. At that point the fill will kick in but, just for the specific block that has a failure. If the next "tranche" of blocks also stall, you'll get the same thing. Stall till the connection times out and the fills then do their thing.

I run 1-2-3 servers all the time and set some fill and some not. I see nothing like this so, I don't think your theory is correct. You make it sound like it's just one group that does this too. That also suggests it's not a logic problem.

"Tranche" a good English word we stole from the froggies. I was a mechanical Engineer now retired after a lifetime of banging and kicking and troubleshooting machinery and later electronic controls.

What makes me think it is a 6.5 problem is that I have tried every beta when it was posted and had the same problem. After each lack of success I then re-installed 6.42 and could download anything from any server with no problems. It was only when beta5 came out I decided to do a little digging and came up with the server configuration problem but I don't know enough about programming to know what is what, that is why I had a guess at the tranche idea.

As for it being a single group, I haven't really tried any of the others - I'll give it a go with some tv shows.

Just had a thought while typing this - wasn't something mentioned after one of the earlier betas about groups in the "Unsorted" gog having to be moved to another gog?

just downloaded 2 tv shows with no problems - they are in a seperate gog though, not in the "unsorted".

Next time it happens, hit the plus icon, look at the files and tell me if you see partial downloads stalled out.

Tried it again this morning and started downloading files whilst "boneless " headers were still downloading. The files downloaded ok until "boneless" finished and disappeared from the download list and then the downloads stalled again. When I expanded the file it was only in the VM colour with no colour from the Easynews fill servers. At a guess I think when there's a connection open to the VM server (boneless headers downloading) the files download ok - when the downloading file tries to continue the connection it stalls for some reason.

As I seem to be the only one having this problem, I removed NB from the PC and have tried to remove every reference to NB I can find in the registry, programme files and App Data folders and then done a clean install of Beta 5. I re-installed the servers and downloaded the groups list and refreshed and emptied out the groups from the "unsorted" folder into various GOG. I kept a copy of my groups in a text file and copied and pasted it into the nbi. file and updated the headers. I've downloaded a couple of files with no problems.

The only strange thing that has happened is that in the "servers" popup window I have gained something titled "Restartpausetime" in the server list with an "X" in a red icon next to it.

I'll monitor some more downloads later/tomorrow and let you know how it goes.

Thanks guys.

The only strange thing that has happened is that in the "servers" popup window I have gained something titled "Restartpausetime" in the server list with an "X" in a red icon next to it.

Suggests some corruption in the NBI Configuration file.

Sort of "Downloading File" by size doesn't appear to be working.
after sorting I see a sequence such as the following.

/8.0
/6.4
/6.8
/9.1

I'll check it out, thanks.

New version of wine today, so I decided to try b5 again.
It crashes/hangs a lot, but will run occasionaly.
I have enabled SSL v2 to speed up my header d/l's and now get
[17:28:56] InterSocket ERROR Error: SSL Negotiation Failed, Host: www.newsbin.com Error: SSL Connection Failed: 6 Error 0 Success.
at startup when it loads. And no header d/l's, a restart of NB and same message but it downloads headers. Velly strange.

That's Newsbin trying to check the motd timestamp so, it knows whether to show the MOTD or not. The connection failure thing is interesting though. I wrote myself a note to investigate.

I got a report of some issues in XP but, I haven't tried it yet. I'll bet wine is pretty close to XP.

HI,

Some strange behavior in b5 when i'm using a nzb (double click) part of it goes to the root download folder en and other parts to the group folder.

example
d:\download
d:\donwload\a.b.boneless\

This was not the case in all the previous versions

Jan

JanJonkman wrote:HI,

Some strange behavior in b5 when i'm using a nzb (double click) part of it goes to the root download folder en and other parts to the group folder.

example
d:\download
d:\donwload\a.b.boneless\

This was not the case in all the previous versions

Jan

Have you checked the contents of the NZB file to see if it specifies a group for each file?

Another issue I have, I can't save files to nzb files.
Happens in b3 & b5, I select the files choose the folder (It forgets the last used folder, which it used to remember) and choose a filename and press save nowt happens, no saved file, no errors.

EDIT:
Looks like something got corrupt. Created a new wine instance, reinstalled NB and it is fine again, so maybe all those hangs have messed with something. Couldn't see anyting obvious in the ini files but it def wasn't using the old folder picker even though it was selected in the options.
Glad it is so easy under wine to install a new M$ folder wouldn't llke to do this under xp again.

mimauk - pretty sure I've discovered the cause of this. Looks like a race condition (meaning I overlooked something). It happens when you have high speed downloads and relatively slow disk IO and high connection counts. The cache empties out and never fills back in. I was able to repro it by, running 40 connections and doing a header import at the same time.

The solution seems to be to check purge chunks on entry and exit.

Thanks Quade.

I'd already give up on it again and gone back to 6.43 for peace of mind. I'll try 6.5 again over the next couple days and let you know.

Thanks again

Quade wrote:
The solution seems to be to check purge chunks on entry and exit.

Installed 6.5 and looked in the "Options" and "Server" settings but cannot find anywhere that lets you "Purge Chunks on exit/entry"

1 - There's no new 6.50 out so what's on the site it what you've already been playing with.

2 - There won't be an option. Just internal changes.

My comment is hot off the presses. I haven't even finished coding yet.

OK - thanks for the update.

HI,
I'm using Newsbin 6.50b5 on a Windows 7 64bits. I've been running out of hard drive space on my c drive. that has never happen before. When I checked why it was because the Chunks folder is huge over 100gigs. it seems like the chunks aren't being purged. is this a bug or am I doing something wrong?

Are you failing a bunch of downloads? When a download finishes cleanly, the chunks are deleted. Then they fail or are removed, they don't. They should be getting purged every 3 days though. I get some lingering chunks but, I know it's because I've failed or removed the download.

Does that seem to see too?

The whole chunk purging thing probably still needs some tuning.

Quade wrote:Are you failing a bunch of downloads? When a download finishes cleanly, the chunks are deleted. Then they fail or are removed, they don't. They should be getting purged every 3 days though. I get some lingering chunks but, I know it's because I've failed or removed the download.

Does that seem to see too?

The whole chunk purging thing probably still needs some tuning.

Yes, I've been removing and assembling incompletes, it weird, all the files only download 99.6/104.4 megs so I needed to assemble the incompletes. and then newsbin takes over and does the repair.

mkrubsack wrote:Installed Beta 5 over Beta 4, and Beta 5 crashes on startup every time I've tried to run it (4 times, then went back to beta 4).

I even deleted the .db3 files in the data folder and restarted. Still crashed on startup.

Beta 4 runs fine. Using WinXP SP3.

Crash at start-up with Beta 5. Beta 4 is fine.
Windows 7 32bits
Single core AMD 2400XP (2.0Ghz)
RAM: 2GB

Crash at start-up with Beta 5. Beta 4 is fine.

Make sure you're only using English language. The beta's don't have language support.

Yes, I've been removing and assembling incompletes, it weird, all the files only download 99.6/104.4 megs so I needed to assemble the incompletes. and then newsbin takes over and does the repair.

How many retries do you have set? Newsbin won't auto-assemble files until the retries expire. I have mine set to 2.

kenr,,

Might have a fix for you in B6. Should be up in the next day or so.

Quade wrote:

Yes, I've been removing and assembling incompletes, it weird, all the files only download 99.6/104.4 megs so I needed to assemble the incompletes. and then newsbin takes over and does the repair.

How many retries do you have set? Newsbin won't auto-assemble files until the retries expire. I have mine set to 2.

I have mine set at 2 too. I think it was all those incompletes on the huge file (50gigs) the chunks added up very fast and didn't get a chance to purge in 3 days. I had to manually purge the chunks. Thanks for your help.

SylvainPro wrote:Crash at start-up with Beta 5. Beta 4 is fine.
Windows 7 32bits
Single core AMD 2400XP (2.0Ghz)
RAM: 2GB

Solve with Beta 6