Wish List Posting

[cheeseontoast]

Is there a problem with posting in the Wish List section?

Wanted to ask if the following could be added:
FTP uses port 21, option to change this value as we have had issues using FTP for years, actinic coming back with FTP errors - looks like after some long investigation that it is our internet provider namely Virgin.

Rackspace changed the port for us to 201 and all was well.

Karl
Movies & Games

[pinbrook]

FTR I've used Virgin for years without any FTP issue

[Duncan Rounding]

Likewise never had a problem with port 21 on NTL/Virgin. Can't see why that should be a problem unless you are using a proxy server for some reason.

[itc]

I've just had some terrible wrestling with an ftp issue, it probably worth posting just for its outright wierdness.

We've been using an optiplex gx745 with xp pro as our actinic machine. Just recently it was decided to move to a much more powerful machine - a Poweredge 1950 /quad cores. We went for xp pro x64 because of the 8 gig memory.

I did the initial install and testing about 4 weeks ago now, and initially all seem to be going well - we went live and everything seem to be ok for the first few hours. Then all of a sudden ftp login errors started occurring during upload but not consistently. It would work sometimes, then throw up errors on the next upload. Wondering if this was connectivity, we dropped back onto the original gx745 and everything was working fine again.

To see if I could get to the bottom I setup a linux box internally on our lan and intially had the occasional ftp error (even more occasional than when live)on that from the 1950. Thinking it was related to the broadcom control suite software and adapter teaming, I removed this and didn't see any more errors. I tested this for a further couple weeks without any kind problems. Time to go live again.

Again the same problems - probably more errors if anything. Assuming because the gx745 worked ok - this must be down to the 1950, I did some googling and came up with a problem with TOE on the adapter - (TCP Offload Engine, -fancy co processing on the network adapters to relieve the CPU). I disabled this - and woohoo - it worked!

Everything seemed fine until 2 days later- this was last thursday it started erroring again. TOE was still disabled and yet the same issues. Thinking this must be due to network issues this time, I contacted our hosts. They assured me nothing had changed and couldn't see any errors in the logs at their end.

I disabled the 2 onboard broadcom NEII nics and swapped them for an intel pro1000. That made no difference either.


I also contacted actinc also and sent the various log files. They suggested to turn off compressed upload (to just use one upload thread instead of the multiple ones when compressed option is enabled). This seemed to work, but was taking an absolute age to do an upload.

The ftp login error was really confusing, because it looked like actinic wasn't closing down ftp sessions when the compressed upload was selected. Knowing that this worked ok on the gx745, I knew there wasn't any intrinsic issues with compressed upload on the server; it had coped fine ever since the feature had been introduced.

Looking on the ftp sessions on the server - multiple ones were opened until the server ran out of sessions then it would throw the login error. It would then take 15 mins for the server to expire the idle sessions to make them available for upload again. I asked our hosts to cut this down to 5 minutes to see if this would help (I had a incredble theory which all seemed to make absolute sense and seemed to fit all the facts lol!) . Ho hum that didn't work either!

Called actinic again and it was suggested to increase the keep alive setting - tried that and didn't help. Looking at the actinic ftp log, it seemed to uploading files but timing out at the end and leaving sessions open and yet no errors being reported at the host. Just clutching at straws I upped the client timeout to 60000us - it worked!

I feel really silly not trying this in the first instance. I excuse my stupidity in knowing that the settings had worked fine on another machine for, well..... months. I can only assume two things going on here, the latency of nics/machine connectivity in combination with inherent changes in wan/host latency.


Sorry for the rambling post. Hopefully it might help someone who finds themselves in a similar situation and this post will apear in any applicable search.

Just now need to mail actsupport and let them know all is now well (rabbits foot, touch wood and co-worker human sacrifice all withstanding )