Steelhead Appliance

[wilkax]

Ok, now I'm really confused...which NO changes made, see what has happened:




Can anybody imagine what could have happened? (the cable swapping theory seems to become true  )

Message Edited by bgilbert on 01-28-2009 04:16 PM

[blancolam]

Hi,

Perhaps I should clarify what I meant by "established connections". By that I mean connections that are either:

1) Connections that were established before the SH service came online. For example, a CIFS connection between the client and server that's already been established before the SH service came online. Obviously, the SH can optimized CIFS traffic but because this connection was already there and hence the SH will not disrupt this connection. (i.e. this particular connection will be in pass-through until it's torn down and re-established)

2) Connections that the SH will pass-through by default. An example of this would be Secure Shell (ssh) traffic or telnet traffic.

Indeed, if you restart the SH service, any *optimized* traffic at the time will be torn down.

Apart from upgrading the software, did anybody swap the cables around on the SH? In other words, the LAN port of the SH connecting to the router and the WAN port connecting to the switch?

Regards,

Blanco

[wilkax]

Hello Blanco,

established connections will not be disrupted when rebooting? Can you please give me a bit more detail about this? AFAIK all connections
are broke down because the optimization service is of course also restarting!?

Apart from this: the connections are all WAN internal...of course some of the passed throughs are to remote locations without a RB and some
are passed through by rules but a large amount of passed throughs are connectionsto a location with a RB...

I'm confused !?

[blancolam]

Hi there,

When you upgraded your SH to 3.0.6 and rebooted them, the SH will not disrupt any established connections by default. Those connections will need to be torn down and re-established before the SH will optimize them.

Another possibility is that there is no SH on the other end of connection (perhaps going out to the Internet?) and hence it's being passed-through.

If you look in the list of connections, what are the connections that are being passed through?

Regards,

Blanco

[wilkax]

Hi there,

for 2 days we're encountering an increasing amount of unoptimized TCP connections (although the SH's are not at their limits). We've
updated our whole infrastructure to 3.0.6.. Another (and perhaps correlating thing) is a Connection Pooling Boost on the central 5010 SH.

Which kind of things can cause this?

Message Edited by bgilbert on 01-28-2009 04:15 PM