PDA

View Full Version : Superhub Another really annoying superhub issue


Milambar
30-03-2011, 06:40
Went to upload a short home video to my server in Germany, 1.1Gig in size..

My server is configured to only accept ssh connections, and on a non-standard port. There is no ftp, etc. Its ssh only. So this means the use of scp or winscp to transfer files.

After 32%, the connection stalled, and my server became unreachable. Fortuntely one of my collegues also has access and he confirmed it was will up and running fine.

I noted the dns was resolving, something was just blocking the port. Rebooted the superhub, and it worked again. Had to do a nasty hack involving rsync -e 'ssh -p$port' --partial --progress.... to "resume" as scp doesnt support resume.

Got to 61% then had to reboot it again.

Got to 98% and had to reboot it again, after which, it finally completed.

Is the superhub designed to block the port like this after so much data being sent, perhaps to discourage filesharing? I was even a good net user, and waited till 3am before starting the upload.

-edit to clarify-
It was only blocking the ssh connection to that particular machine. ssh to my other 3 machines all still worked fine (they are not mine really, but belong to the company I work for, free webspace is a bonus of the job).

Chrysalis
30-03-2011, 08:30
I suspect the session may have simply timed out as the superhub has a very agressive timeout value set which cannot be tuned in the gui. So in that situation I would try to force keepalive packets with either the scp client or sshd server to more than every 60 seconds and see if that resolves the issue.

craigj2k12
30-03-2011, 12:57
is this a problem with multi-cast?

foddy
30-03-2011, 20:07
is this a problem with multi-cast?
I didn't know VM even supported multicast!

Milambar
30-03-2011, 20:23
I suspect the session may have simply timed out as the superhub has a very agressive timeout value set which cannot be tuned in the gui. So in that situation I would try to force keepalive packets with either the scp client or sshd server to more than every 60 seconds and see if that resolves the issue.

While an SCP transaction is active, data is being sent, bidirectionally. That is, Im sending data to the server, and its replying with ACKnowlegements. Therefore it will never time out, as the session is not idle.

Even so, I altered sshd_config a long time ago, well over a month ago, to send TCPKeepAlive's every 59 seconds.

In this case, it seems that the suberhubbing superhub (yes, I used superhubbing as a cussword) actively blocked the connection itself, after a set amount of data or time.

Chrysalis
30-03-2011, 21:38
59 is probably pushing a bit close to it, maybe set to 30s or so.

Is the firewall and flood protection stuff disabled in the superhub?