ioFTPD v7.3.3 Released (Beta)

[pion]

Caching the DNS lookup from a task manually added to the scheduler should be plenty. Entrybncs are not always on static ips, often cases where cuBNC is involved.

[o_dog]

i'm still having problem with the lockup of a transfer and then denying transfers until you login again. Using latest version.

[Yil]

o_dog: Check your logs. The example pion posted clearly shows the FTP client at fault. It requested a transfer via a PORT command and after 10 seconds the 2nd site timed it out and the client just starting issuing PORT commands to retry without aborting the still running command. After an additional 10 seconds ioFTPD goes ahead and declares it failed and then things would start working again... If you wait 30 seconds does it start working again for you? It sounds like it doesn't, but just checking. Some logs would be useful showing the problem, send them to the FTP crash site if you can cut/paste an example of the problem. PM me if you need the login info.

I think I got openSSL working. I can login, enter commands, transfer files, etc. Need to finish up a few things but looking good...

[o_dog]

pm'd you about it, but i can't use this as it stands now. I can't transfer files with ioFTPD right now which is kinda dealbreaking for me. plz try to fix this one fast

[Yil]

PM'd. Can you confirm from your client logs that this happens only when FXPing? And does it only happen with drftpd as the other side like in pion's example? The FTP client probably handles things a bit differently with the PRET command. My money is still on whatever FTP program you are using... Any chance you upgraded/changed it recently?

[o_dog]

no, it's gllftpd. I don't go near drftpd. And no, i havn't upgraded recently
or any such thing, except for ioftpd, gonna check for the logs of the ftp client.

[pion]

Yil; I also got the same behaviour in direct download, with client sending ABOR (unlike fxp, where it didn't send ABOR) - where all transfers got blocked. PORT, PASV, fxp, direct. But the fact that io was blocking both pasv and port for all sessions, including new login, and the fact that io locked up a few minutes later denying all connections makes me assume that it's symptoms of the lockup bug (or the cause), rather than a separate incident.

[monk-]

Originally Posted by pion

BNC_HOST_1 = 192.168.0.* <-- doesn't work, sais "doesnt resolve" in log

what about using wildcard for iprange, can u support this too, yil?

[Yil]

monk-: If you need IP wildcards for the BNC host address (reverse resolved names probably as well), I can do that. And fully specified names for looked up via site rehash or via a new scheduler event sounds good too so it can handle dynamic DNS names if needed.