Zeus User Community

Hello

I'm setting up a test lab of TMG 2010 in an array configuration of two servers to be used as just generic http/s proxy servers. I then use the ZXTM as the external load balancer that the clients connect to to give it fault tolerance.

I've had to set the virtual server to generic client first so my clients browsers can connect to SSL enabled websites.

The only downside to this is now I cannot apply traffic script rules, namley x-forwarded-for so the logging mechanism on TMG can then interpret the header as the originating request.

Can anyone think of a way around this?

I'm having some trouble importing a wildcard cert. I did this a few months ago for another domain with no issues.

I keep getting this message but cant work out as to why

Failed to import Certificate (Certificate does not match private key)

Anyone come across this before?

It records it as the gzip size if the client accepts the compression.

I've seen three memory issues over the years when using a hardware appliance

1) LCD driver for the front control panel, dont think your suffering from this. We had to disable this. Might be vendor specific (we use JetNexus)
    * pgrep lcd (this will tell if you its running with a PID)
    * rm -f /etc/rcS.d/S41usb-display (This will remove the driver from being used)
    * kill -9 PID (This will then kill off the process, replace PID with the number in the first step)
2) There was some sort of memory leak in the old 5.1 Kernel when using BW limiting. We solved this by upgrading to 6.0 (http://bugzilla.kernel.org/show_bug.cgi?id=8971) We are currently on 6.0r4
3) Some issues with a bug in traffic script not releasing memory.

Your best bet is to log a ticket with Zeus and submit your TSR from your boxes.

Sorry should have written an update.

Got a reply from Zeus, we'll be upgrading to version 6 shortly

“This isn't working because you need SOAP API permissions enabled, only the admin group has that option on version 5. On version 6 you can setup any group to have full permissions to use it.”

For one of our new web applications (Developed in asp.net WCF) they are going to be using Net TCP for some of the web services that talk to each other internally (http://msdn.microsoft.com/en-us/library/ms734772.aspx) I dont know much about Net TCP as I've only just be introduced to the project for the web application. We currently have this configured in DEV and working but I'm just wondering if the is the best configuration or under load we might start to see some issues. I'll try and explain below.

So we have 1 virtual server and lets say 3 pools. The default pool is the Web UI layer and then through traffic script depending on what path is requested that then gets pushed through to one of the other 2 pools.


Virtual Server Listens on HTTP for 192.168.1.1
Default Pool (Web UI Layer) going to IIS 7.0 instance on Server 1 Port 80 192.168.1.2
Traffic script intercepts any requests for /error-logging/ & /db-query/ and passes them to the respective pools.
Error Logging Web Service going to IIS 7.0 instance on Server 2 Net TCP Port 5085 92.168.1.3
DB Query Service going to IIS 7.0 instance on Server 3 Net TCP Port 5085 92.168.1.4

Now lets say I configure the UI layer via my web.config to report any errors to "net.tcp://192.168.1.1/error/logging"

Obviously that will connect to the Virtual Server on port 80 and then pass through the request to Server 2 and this does seem to work.

I'm just a bit concerned that its a "net tcp" protocol rather than an "http" protocol and this may have some kind of an effect as the ZXTM is expecting regular http traffic or on the backend nodes.

Has anyone else used net tcp webservices in nodes and virtual servers with ZXTM 5.1r2 before.

Any comments, questions or dont be stupid! would be appreciated

Turns out it was a problem with IPV6 connectivity to the ja.net servers. I used the IP's instead of hostnames and it worked fine.