ERROR - CNNTPServerWorker - internal crash

[shiai]

With the last RCs (now build 1528) i get nearly all the time "ERROR - CNNTPServerWorker - internal crash" and the download stops. Restarting doesn't help.
It happens while downloading nothing spectacular ( This also happened with the last RCs.

[itimpi]

As far as I know, you are the only person who has reported this, so is there anything specific to your system that might be a contributing factor?
For instance:
- which version of Windows do you use?
- what anti-virus/firewall software?
- do you use the 32-bit or 64-bit version of Newsbin?
Unless Quade can get enough information to give him a chance of reproducing the issue then it is much less likely that it will get fixed.

It might also be worth doing some checks that the trigger is not hardware related. Running a RAM check using something like memtest86+ seems a good idea that is easy to do as if you have some RAM that is borderline it might cause this type of symptom. A hard disk check using the disk vendors diagnostic software might also be a good idea.

[shiai]

No special System:
Win7 32 bit with 32bit version of newsbin, only win7s internal firewall and MS Essentials ... as usual the last months. There is nothing special running on this system, as it is my "download-only"-computer.
as i mentioned, these problems occured since the last release candidates, before there were no such errors.
The strange thing is, that it does not happen with other files - at least till now as far as i have seen.
Also when i try to redownload (deleting the files and then bypass filters ctrl+y), the same error occurs again.
Forgot to mention: All Downloads are stopped then, not only the "problematic ones".
When i delete the "problematic ones" from the queue, and i restart newsbin, the download works.

[Quade]

Check how much RAM you're using when this happens. I actually induced an error like this the other day. When I checked, Newsbin was using 3 GB of RAM which on a 32 bit system means I'd run out. I was testing loading up some massive groups. The loading worked fine but, 32 bit programs can't use more than 3 Gb of RAM. It didn't leave enough RAM for downloads to work so, the downloader crashed.

[pinaar]

I'm getting the same thing under Windows XP.
It started around 10/9/11.
My computer has 2 GB RAM, Celeron CPU 2.93 GHz, 11 GB free on the 100 GB C drive.
I only recall adding my wife's account to the computer so she could share the printer from her laptop, which I already could from my laptop.

I've restarted the computer about 3x since then.
NewsBin Pro 6.11 Build 1342 will download the group index,since it shows a greater than zero rate on the status bar for a few seconds, then goes to zero. I just tried downloading a small 200 KB file and it won't.
[added this a few minutes later - NewsBin is consuming 92 MB, Memory Total=2 GB, Available=1.1 GB, Sys Cache=0.95 GB, Kernal Total=80 MB, Paged=67 MB, Nonpaged=13 MB.Handles=12290, Threads=581, Processes=42.
Commit Charge: Total=2549368, Limit=4054960, Peak=4054864.

I just applied "Automatic Updates" and will reboot. I'll see what happens.]

- Art

[Quade]

Have you looked in the logging tab? Sounds like you're not getting a connection to the server at all. A little bit of data rate is often just error messages. I'd suggest that your account has lapsed with the news server.

[pinaar]

Thank you for the quick reply Quade. I just logged onto my GigaNews account and it says it's active, but that I have no current connections!
I'll reboot and resume the investigation. - Art

[pinaar]

Working fine after this reboot (Automatic Updates applied).
The logging tab showed,
[21:39:14] HIGH Download Engine GN - INT_OpenItem -DL_DI - Failed to retrieve Article from Server::GN
[21:39:14] HIGH NNTPSocket - Socket HandleError - Server: news.giganews.com SendCommand - Server Reports that a Post isn't on the server: 430 no such article
...
The pop-up warning thingy showed,
[21:41:53] ERROR - AssembleFiles: No Filename - Filename from subject:(xxxx) [01/29] - "xxxx.S01E06.720p.HDTV.DD5.1.x264-CtrlHD.par2" yEnc
[21:41:53] ERROR - INT_AssembleFile: Assembly Failed (xxxx) [01/29] - "xxxx.S01E06.720p.HDTV.DD5.1.x264-CtrlHD.par2" yEnc
[21:41:53] ERROR - AssembleFiles: No Filename - Filename from subject:(xxxx) [21/29] - "xxxx.S01E06.720p.HDTV.DD5.1.x264-CtrlHD.sfv" yEnc
...
The articles above were a) modified to protect the innocent, b) came from an NZB that was over 300 days old, so it's quite likely that the missing files really are and are NEVER to be seen again.

Thank for the quick response, but I think it was a transient error due in part to missing deleted files (I'll delete the target downloads while NewsBin is tweaking them when I'm done with them. Perhaps the proper process is to tell NewsBin I'm done with them first.).

Thanks again,
Art

[pinaar]

Downloaded several articles, but it ultimately ran out of memory.
It first had a bunch of,
[23:05:30] ERROR - FileIOThread - internal crash
then,
[23:29:19] ERROR - CNNTPServerWorker - internal crash
I restarted NewsBin and opened 10 groups at once again, while the NZB continued trying do download the 300+ day old articles.

- Art

[Quade]

[21:41:53] ERROR - AssembleFiles: No Filename - Filename from subject:(xxxx) [01/29] - "xxxx.S01E06.720p.HDTV.DD5.1.x264-CtrlHD.par2" yEnc
[21:41:53] ERROR - INT_AssembleFile: Assembly Failed (xxxx) [01/29] - "xxxx.S01E06.720p.HDTV.DD5.1.x264-CtrlHD.par2" yEnc
[21:41:53] ERROR - AssembleFiles: No Filename - Filename from subject:(xxxx) [21/29] - "xxxx.S01E06.720p.HDTV.DD5.1.x264-CtrlHD.sfv" yEnc

This is more of a warning than an error. It means Newsbin couldn't find the filename. It might also mean none of the file was able to download and Newsbin is trying to assemble what it's got.

I restarted NewsBin and opened 10 groups at once again, while the NZB continued trying do download the 300+ day old article

I'm unclear what you're saying here. I think you're saying "if I consume enough RAM bad things happen".

[color]

Have the same ERROR - CNNTPServerWorker - internal crash messages since one of the lasts V6 versions.
In the evening I start a download queue with about 200GB which will not run through.
Got masses of ERROR - CNNTPServerWorker - internal crash messages in the morning and also autopar didnt work fine sometimes.
There are empty directories with no file in and sometimes only a par2-file is there.

Unfortunatly I have a 32bit Win7 netbook with icore5 Processor 3GB RAM and a 100MBit Internetconnection via a cableprovider.
With the old V5 and earlier V6 versions I did not have this problems. So something must changed.
Maybe you can fix the problem for 32Bit version or do you have some workarounds for poor 32bit Windows.

Thanks

color

[Quade]

Try this. Exit Newsbin then, add this entry to the configuration file.

[SETTINGS]
ChunkCacheSize=50

Using wordpad. Make a backup of the NBI first.

This will reduce RAM usage by as much as 100 megs or more.

[pinaar]

Thanks, Quade. I added the,
ChunkCacheSize=50
you posted for a previous poster.
Yes, on mine, it's pretty much as you said,
I think you're saying "if I consume enough RAM bad things happen".
I'd always had this issue with newsreaders. If I try to download large newsgroups article lists or keep multiple newsgroups at on my tabs for reviewing them at once, readers run out of memory and or hit parameter limits.
In this case, the dialog pop-up titled, "Newsbin50" displayed the message "Out of memory" every time I click on the "x" to kill the main interface window. I used Task Manager to kill newsbinpro.exe which was at 1,421,804 KB (system only has 2GB RAM running Windows XP 32-bit).

I haven't seen the crash since I did the above NBI change, but I've been running it for 20 minutes, downloading still.
An unrelated error I'm seeing is,

[08:25:42] HIGH Decode Block - DecodeYenc - Bad YEnc Format: xxx.s01e05-french LD -HDTV bettyboop et bruno.avi.038


Shows up for some old downloads. It may be that they're corrupted or possibly using a non-conformant yEnc encoding, but I thought I'd mention it.
- Art

[Quade]

system only has 2GB RAM running Windows XP 32-bit).

If you followed the procedure to set the /3Gb switch, this problem would probably go away. It's not a problem with RAM because the pagefile gives you near unlimited RAM. The problem is address space. Even if you have 4 Gb of RAM, Newsbin can't use more than 2 because of Windows 32 bit limitations. The /3gb switch gives 32 bit applications access to another 1 Gb of address space.


http://dwf.blogs.com/beyond_the_paper/2 ... vista.html

Is a procedure for turning on the /3gb switch. Newsbin is large address aware so, It can use up to 3gb in a win32 environment.

[marcus1260]

Hi Quade,
I am having the same issue on my Win 7 (32) machine with 4 GB RAM installed. Below are the log messages.

[08:36:02] HIGH - High Speed File Creation Mode: Enabled
[08:36:16] HIGH InterSocket - SSL Zlib Compression Added
[09:25:09] ERROR - CNNTPServerWorker - internal crash

Will the solution you suggested work for me?

Oh my version is 6.11 build 1342

[Quade]

I don't know. You can only try and see.

[UlrichKliegis]

Quade and folks,
I can report that I today also saw this error (nb 6.11, XP pro, Amd Athlon 64, 1 GB RAM, gigatons of disk space) here when I saw frequent interruptions of the data stream. In my case, for historical reasons, port from an (yet!) older machine, I had the .nbi file in a non-standard location, which was still pointed at in the nb setup. Well, I added the chunkcachesize parameter to the settings section, saved, and started again. No traffic.

It quickly turned out that this (extratopic) nbi file still contained the access data of my previous provider (changed in july 2011). NB had worked woth those new data for 4 months already. Correcting this in NB brought back the traffic - but only at 5 Mb/s, previously 12. The traffic limiter box was unchecked. Checking it on, setting 12 Mb/s, unchecking it again and restarting NB brought back the higher speed. And no more crash errors! Image link not allowed for unregistered users

I'll touch a different topic (UnRxxxxx.tmp folders everywhere Image link not allowed for unregistered users ) in a separate post.

Cheers,
U.

[haileris]

Just to note that on the latest beta build I am getting this quite a bit (6.30b2). Sometimes it recovers but today it just spurted it out about 15 times into the event log then fell over. Getting lots of probs with this build, sometimes newsbin just disappears (CTD). Running in a Win2k8R2 x64 VM with 4 GB. A duplicate of this running flawlessly (for the most part anyway) on 6.20.

Cheers

[Quade]

I tracked this down to header downloads. The internal crash thing. It's an issue in B2. What happens is a logging buffer gets too large.

I think it happens if it's run long enough which probably explains why I didn't see it in normal testing. B3 fixes it. I don't think B1 had this. Looks like I broke it during static code analysis. Made a number unsigned and it broke.

[haileris]

I tracked this down to header downloads. The internal crash thing. It's an issue in B2. What happens is a logging buffer gets too large.

I think it happens if it's run long enough which probably explains why I didn't see it in normal testing. B3 fixes it. I don't think B1 had this. Looks like I broke it during static code analysis. Made a number unsigned and it broke.

Super, thanks for your quick response [and honestly ]

[BFHammer]

Files that help describe the problem:
C:\Users\David\AppData\Local\Temp\WERAAD7.tmp.WERInternalMetadata.xml
C:\Users\David\AppData\Local\Temp\WERC27D.tmp.appcompat.txt
C:\Users\David\AppData\Local\Temp\WERC2FB.tmp.mdmp
from the windows crash

[08:28:38] ERROR - FileIOThread - internal crash

It runs for about a minute then crashes out and windows closes it down. I'm on 64 bit with 4g Ram win 7.

I deleted everything in the download folder and now it's been functioning for a few minutes without a crash.

[Moondawgie]

I tracked this down to header downloads. The internal crash thing. It's an issue in B2. What happens is a logging buffer gets too large.

I think it happens if it's run long enough which probably explains why I didn't see it in normal testing. B3 fixes it. I don't think B1 had this. Looks like I broke it during static code analysis. Made a number unsigned and it broke.

[Thanks Button]
(http://www.phpbb.com/customise/db/mod/thanks_for_posts/)

[super_ius]

Tried all the workarounds presented on this thread.
Giving up--downgrading back to 6.11 is the only sol'n that works for me.

UPDATE: I tried 6.30RC1 on a whim, and it is working just fine for me.