Beta 5.5.0

[darkone]

All:
-New: Fibers, ini entry under [threads]: Worker_Fibers = #.. use 2-5x the amount of worker threads you have)
-New: Executables (scripts) now have limited amount of time to run without output (first output is expected within 2minutes, second within 30minutes)
-Change: Message threads removed (unified with worker threads)
-Fix: Rehash

No new commands.

[Pichento]

using site invite (ioA 1.0.5) will disconnect you from site using ioFTPD 5.5.0

prolly a bug with ioA but still very annoying...

Be warned

[Pichento]

First race and everybody was disconnected after 10 secs.

ftpd crashed completely

[Pichento]

maybe itcl is somehow broken in 5.5.0?

reverted back to 5.4.3 for now. rock stable.

i tried checking systemerror.log but nothing there.

5 x Reports (in 60 secs) from event viewer:

Faulting application ioFTPD.exe, version 0.0.0.0, faulting module ioFTPD.exe, version 0.0.0.0, fault address 0x0003e56b.

[Mouton]

-New: Executables (scripts) now have limited amount of time to run without output (first output is expected within 2minutes, second within 30minutes)

this exclude !detach'ed scripts, does it ?

could still be a problem for pre scripts that move stuff to a different drive... no output is done while transfering the files... and transfering 4gb from one drive to another could take more than 30 secs...
script would be required to ouput a kinda a %-bar ..?

[darkone]

Yep, it's only for those that are bound to client (to prevent stalling of connections)

[Ren]

dies juring race

address, 0x000000000043e5aa

[darkone]

Originally posted by Pichento
maybe itcl is somehow broken in 5.5.0?

reverted back to 5.4.3 for now. rock stable.

i tried checking systemerror.log but nothing there.

5 x Reports (in 60 secs) from event viewer:

Faulting application ioFTPD.exe, version 0.0.0.0, faulting module ioFTPD.exe, version 0.0.0.0, fault address 0x0003e56b.

As usual, email me your whole configuration.

[Luke]

Originally posted by darkone
As usual, email me your whole configuration.

Could you please add that "refreshdir whileuploading" thingy tuff made as server default ?

0-byte thingy

Thanks

[Mouton]

haha
directory cache are good, not bad.
scripters need to flush the cache when they change something outside io, is all...
can be done in tcl and shmem.

as for knowing the size of currently uploaded files... i don't really see the usefulness, except curiosity. and even then, a SITE FLUSH itcl command that does what tuff thingy does would be enough.

[darkone]

Yap, most unix geeks are drooling over built-in list commands in ftp daemons... (executing 'ls' creates too much overhead) Directorycaching is even neater (it removes nearly all overhead)

[Mouton]

tss... geeks...
we don't take kindly of those around here!

[Pichento]

5.5.3 seems to have fixed all my probs. Darkone, coould you inform us what was wrong?

Curious.

[darkone]

Buffer overflow.. used wrong variable when cycling the buffer (in pipe redirector)

[Pichento]

Heia Darkone.

Good work with eliminating bugs in 5.5.x core so far.

5.5.3 seemed pretty stable but after 14 hours of uptime, it suddently crashed this morning. Was probably due to site invite from ioA, but can't say for sure.

I also got a lot of errors from schedueler (events return errors now), but at first hand they do not seem to crash ftpd. I have reverted back to 5.4.3r for now as stability is paramount.

Keep up the good work.