Connection problems

[Sally W.]

Customer has the latest GDC and runs via Citrix and a tunneled VPN.  Server is running Solaris 10 and used to be running Solaris 8.

Since the server was upgraded they have been unable to connect so I visited the site yesterday.

I made a new connection that uses telnet and displays the terminal window, and has nothing in the command box.  I used it to log in as 1) a user with a .profile and 2) a user without a .profile

In each case I got a message that "<logname> could not be executed".

When I added @FGLKSH to the command box $FGLSERVER wasn't set.

If I ran the product setup script and then appended the correct offset to $FGLSERVER the product ran so the firewalls don't seem to be a problem.

Normally the command box contains: @FGLKSH; . /setupscript; exec go

Most non-Citrix users have the second two commands in their .profile.

Another remote Citrix / VPN tunnel site with the same shortcut setup is working just fine, all the users on the same network as the server are working, no problems for me from home (VPN but no Citrix) and no problems for the non-Citrix remote sites.  This site has 6 Citrix servers with the same problem on each, which in my view makes corruption of the GDC installation most unlikely.

Has anyone ever had anything similar to this? 

What I haven't been able to find out is what port the @FGLKSH works over (6400 + offset?) but I spoke to the people that run the site's firewall and they said that all ports are open in all directions over the VPN, and they couldn't see any obvious traffic problems.

Yes I have a support call, but someone here might know the answer or have some useful input and if I don't ask they will never know we have this problem.

[.]

Sally,

First question: which GDC version are you using ? If you're running 2.32, if you record a log file (same way as for debugging a "normal" GDC issue), some information will be printed there when running your shortcut.
This may help indicating you what's going wrong.

Second question: how does GDC connects to your Solaris server (ssh or telnet)  ? If ssh, are you using "Start command in a new shell option" or not ?
If you're using telnet or ssh+start command in a new shell, you need to ensure the connection strings are still valid for the new server.

For instance, if your Solaris 8 server was displaying "user password:" for password query and "Last login:" when connected, and you were using "login:" as "send command" trigger, the connection will fail if now the servers shows "user login:" to query the password.
You may also check the connection string priority and options ("only once", "ignore remaining strings"...) which may influence the connection.

From what you describe, it looks like the command is never sent, which usually corresponds to a connection string issue ; so first step is to check them, look a the log to see if something is relevant. If you're unfamiliar with these, send the config.xml file and the log file to your support center, we'll have a look.

Regards,
Pierre-Nicolas

[Sally W.]

Hi

Thanks for replying.

This is version 2.32 of the GDC and it uses telnet to connect.

However I've realised that the software was upgraded on the Citrix server but the shortcuts are from the earlier version (I've no idea what that was) with the result that the new FGLAUTOPORT string isn't present.  Possibly there are some other changes as well.  So, I've emailed the site with screen shot by screen shot instructions for creating a new connection - the one I used to test yesterday was duplicated so had the same strings as whatever the earlier version of the GDC created.

I'm not confident but it might be an interaction between the upgraded shortcut and Solaris 10 and creating a new connection is certainly well worth doing.  If that doesn't work then yes, I will read up on recording a log file and hope that shows me where the problem is.

Sally

[.]

Hi,

There's just something you should clarify: autoport is only working for SSH (when using auto port forwarding). It makes no sense when using telnet.

Regards,
Pierre-Nicolas

[Sally W.]

OK.  However, using the new connection has moved things forward a bit - it no longer gives the 'cannot execute <logname>' message.

Instead we have a different one, I've attached the screen shot of the terminal screen the user sent me.  She logged in as a user who has the Korn shell - /bin/ksh.

[Sally W.]

We think we have fixed this.  When the new connection was created a colon rather than a semi-colon was accidentally used after @FGLKSH.  It's visible if I look closely at the screen shot.  However the user has fixed it and it seems to be working!