View Single Post
Old 06-14-2008, 10:49 AM  
thomas74
Junior Member
 
Join Date: Dec 2007
Posts: 14
Default

Quote:
Originally Posted by Yil View Post
thomas: Describe exactly what you see happening. Can you issue "site crashnow" and does it work? If you try to connect to the server immediately after you see it acting wierd do you at least get as far as the login prompt, an accepted connection, or does the connection just time out?
1) I send a release to the server containing two cds(two folders) and a sample(one folder). The dirs are always created successfully, and if I manage to get the SFV in, it seems like all files in the directory will transfer. It would seem that the server locks up during the directory changes. Let's say CD1 was transferred successfully and I start uploading the sample. I make the directory SAMPLE, but rather than completing the transfer I get a message similar to "no connection could be made"; and the same message is repeated until the entire queue transferred unsuccesfully. When the bug already have occured and I try to upload something, the transfer will just stall for minutes, hours or days, until I stop it -- no errors. I am of the impression that this bug occurs only during directory changes.



2) I can issue SITE CRASHNOW, but I don't see any dump being created. I see a crashlog being created, but I'm not sure if it's due to my ctrl+alt+del ending of ioFTPD or me issuing SITE CRASHNOW. From the crashlog:

Quote:
Fri Jun 13 23:08:08 2008 - ioFTPD v6.6.0
Unhandled exception: Access Violation (0xC0000005)
Address: 0x0042BFDD [attempting to write data to 0x00000000]
PID=7044, PATH=C:\ioFTPD\system\ioFTPD.exe
Thread ID: 4836

System information:
Processor #0 Name: Intel(R) Core(TM)2 Duo CPU T5450 @ 1.66GHz
Processor #0 Identifier: x86 Family 6 Model 15 Stepping 13
Processor #1 Name: Intel(R) Core(TM)2 Duo CPU T5450 @ 1.66GHz
Processor #1 Identifier: x86 Family 6 Model 15 Stepping 13
OS: Windows 6.0 (build 6000)
Registry: Windows Vista (TM) Home Premium
Decoded: Vista Home Media Center Edition
Page size: 4096
Notice: Unable to acquire Debug Rights
3) Usually I am able to login after the lockup has occured, I can also issue various commands. Other times again, I'm not able to log in at all; the server won't connect to the client at all. I haven't been able to see any pattern.


Quote:
Originally Posted by Yil View Post
And describe your hardware. I don't think you have a hardware problem, but I was able to see this bug on multi-processors in the past more often so that's useful to know...
Yes, as you can see above it's a core2duo with 2 gigabyte or RAM.


Quote:
Originally Posted by Yil View Post
Microsoft offers a tool called app-verifier. It's freeware. Try downloading that and setting it up to monitor ioFTPD. It will make ioFTPD use a lot of memory and be a bit slower but it can catch a lot of issues. If it finds any let me know. There are 1-2 false positives, but for the most part if it complains it's something that needs fixing. You might need to download MS Tools for Debugging in order to get windbg to create the stack traces app-verifier requests to assist in debugging, so grab that as well... This is the single best way to help solve this issue.

Ordinarily I'd say just send me a crashnow dump so I can see what's going on, but because the process is so badly messed up that won't work. You might be able to use the vista debug/memdump or whatever option from task manager to create a minidump of the process. I can use that to verify it's the same problem I've been trying to track down. This is actually far less useful than the app-verifier tracking because that might catch the problem when it occurs and not when it shows up later.
Alright, I will try installing app-verifier to see if anything can be found.

Quote:
Oh, as far as last version without this issue. I have no idea and believe me I wish I did... I'm pretty convinced it's been there forever. It's the fact that we started getting multi-processor and faster machines that is making it show up now...

Assuming you aren't running recent versions of ioNiNJA/ioSFV you could always try going all the way back to 585r just to prove it didn't occur there if it really does happen every week or so to you...
Hehe, could be worth the try. This bug occurs several times a week, and if I wanted I'm sure I could make it occur several times a day. Trouble however, is that I don't have physical nor complete remote access to the vista machine.
thomas74 is offline   Reply With Quote