FlashFXP Forums - View Single Post

Yil · 12-04-2011, 10:20 PM

7.7.x releases are light years ahead of previous releases in terms of stability.

7.7.1 got rid of the OpenSSL memory corruption issues caused by poor OpenSSL documentation. I still can't believe they don't provided a default locking mechanism for thread compiled libraries...

7.7.0 includes fixes for a large number of memory issues. I tested every built-in site command with a number of different arguments as well as a variety of other things. It also includes a simple corruption detector to check the internal reference counted allocations. To date nobody has shown any corruption detected log entries which I can only take to be a good thing

On the other hand the annoying loader lock problem still exists for some people, but as long time users will know by now it's a subtle problem and very tricky to debug given that it's a windows internal lock and ioFTPD has no control over it directly and that it breaks nearly everything when it locks up. The switch to OpenSSL really helped reduce the problem, but I still don't know what the real cause is, only that some things make it worse or better which helps point me to other things to check.

I can say that I have yet to see a lockup/crash myself on the new versions personally even on servers with 3-5 average transfers at a time and 20 online users. I have seen a machine burning dvds or watching a movie timeout a service and force a restart but that's because it was a slow machine and the server didn't get CPU time to run... I presume those with problems must have lots of users online so hopefully you've set the number of worker threads high enough that they aren't being created/destroyed as I know that helps trigger the bug.

I've gotten no time to play with anything for a while, so you'll just have to make do for the time being...

pion: You're using nxMyDB right? Using the customized version of that I made for you I can't see why that would be an issue, but I'd be curious if you still have the problem if you could find a way to test your setup without it since that's a pretty sizable difference between what you are running and what most people are using.

12-04-2011, 10:20 PM
Yil Too much time... Join Date: May 2005 Posts: 1,194	7.7.x releases are light years ahead of previous releases in terms of stability. 7.7.1 got rid of the OpenSSL memory corruption issues caused by poor OpenSSL documentation. I still can't believe they don't provided a default locking mechanism for thread compiled libraries... 7.7.0 includes fixes for a large number of memory issues. I tested every built-in site command with a number of different arguments as well as a variety of other things. It also includes a simple corruption detector to check the internal reference counted allocations. To date nobody has shown any corruption detected log entries which I can only take to be a good thing On the other hand the annoying loader lock problem still exists for some people, but as long time users will know by now it's a subtle problem and very tricky to debug given that it's a windows internal lock and ioFTPD has no control over it directly and that it breaks nearly everything when it locks up. The switch to OpenSSL really helped reduce the problem, but I still don't know what the real cause is, only that some things make it worse or better which helps point me to other things to check. I can say that I have yet to see a lockup/crash myself on the new versions personally even on servers with 3-5 average transfers at a time and 20 online users. I have seen a machine burning dvds or watching a movie timeout a service and force a restart but that's because it was a slow machine and the server didn't get CPU time to run... I presume those with problems must have lots of users online so hopefully you've set the number of worker threads high enough that they aren't being created/destroyed as I know that helps trigger the bug. I've gotten no time to play with anything for a while, so you'll just have to make do for the time being... pion: You're using nxMyDB right? Using the customized version of that I made for you I can't see why that would be an issue, but I'd be curious if you still have the problem if you could find a way to test your setup without it since that's a pretty sizable difference between what you are running and what most people are using.