Virtualization

Stefan,

The first things that spring to mind are;

1) Corruption of the .pkg files - this would fit with the fact that a repair resolves the issue. How is your SG Client user data configured? Is it on a network share?

2) It may be that your client cannot connect to the package - your URL is pointing at NLB51.fun.bbb.org - this sounds like a Loadbalancer - perhaps there's a problem there (hence the intermittent nature of the problem) - can you try changing the path to point
to a specific VAS to see if the problem persists?

Cheers,
Danny

The citrix servers are using roaming profiles. The users are working on local profiles that are saved on a network share when they log out.
Doing "repair" or "sftmime /remove obj:app" and logging off and on logged in as the user fixes the problem sometimes.
But after a while that doesnt work anymore, like there is something going wrong with the "application settings" too, not only the user profiles.

The error only happens on the Citrix servers, not on desktops.

You are correct The NLB51.fun.bbb.org is a MS load balancer pointing at 2 VAS.
Bypassing the load ballancer might be worth trying.


/Stefan

With load-balancing, make sure that the affinity between connection to 554 and then to two high-ports are retained. In case of NLB, this means either "Class C" or "Single" affinity for rules (TCP 554 + TCP 49152-65535).

/Kalle

Hi, haven´t looked into the NLB config /bypass yet...
If there is a problem there shouldn´t the desktops be affected as well?
Here´s a verbose log of a "launch failed":


[10/18/2007 08:21:30.865 SWAP VRB] {hap=10A9:app=Kodak_imaging 5:tid=1178:usr=nisse}
Starting launch thread
[10/18/2007 08:21:30.865 SWAP VRB] {hap=10A9:app=Kodak_imaging 5:tid=2504:usr=nisse}
Running in launch thread
[10/18/2007 08:21:30.865 SWAP VRB] {tid=2504:usr=nisse}
PackageData::GetUrl()
Original: rtsp://%SFT_SOFTGRIDSERVER%:554/00000127/00000127.sft
Returning: rtsp://NLB51.fun.bbb.org:554/00000127/00000127.sft
[10/18/2007 08:21:30.865 SWAP VRB] {hap=10A9:app=Kodak_imaging 5:tid=2504:usr=nisse}
Calling FSI->Connect for url rtsp://NLB51.fun.bbb.org:554/00000127/00000127.sft
[10/18/2007 08:21:30.865 JGSW VRB] {hap=10A9:app=Kodak_imaging 5:tid=2504:usr=nisse}
Connect(url=rtsp://NLB51.fun.bbb.org:554/00000127/00000127.sft)
[10/18/2007 08:21:30.865 JGSW VRB] {hap=10A9:app=Kodak_imaging 5:tid=2504:usr=nisse}
Using Local Appdata Directory: C:\Documents and Settings\nisse\Local Settings\Application Data\SoftGrid Client
[10/18/2007 08:21:30.865 INOT VRB] {tid=2504:usr=nisse}
IN: SWFsdIoManager::Connect
[10/18/2007 08:21:31.193 INOT VRB] {tid=2504:usr=nisse}
OUT: SWFsdIoManager::Connect
[10/18/2007 08:21:31.193 JGSW VRB] {hap=10A9:app=Kodak_imaging 5:tid=2504:usr=nisse}
Elapsed time for SWFsdIoManager::Connect(): 0.328 seconds
[10/18/2007 08:21:31.193 JGSW ERR] {hap=10A9:app=Kodak_imaging 5:tid=2504:usr=nisse}
The SoftGrid client could not connect to stream URL 'rtsp://NLB51.fun.bbb.org:554/00000127/00000127.sft' (FS status: 16D1380A-0000E005)
[10/18/2007 08:21:31.193 SWAP ERR] {hap=10A9:app=Kodak_imaging 5:tid=2504:usr=nisse}
The client was unable to connect to a SoftGrid server (rc 16D1380A-0000E005)

/Stefan

I have solved the case now.
this is what i did:

1. moved the sg clients %appdata% folder from the users profile to their home folder.
this made things alot better but did not make the problem go away completely.

2. configured the sg client to never show the blue status bar and icon down by the clock.
This made the problem go away completely. (if i had done this first maybe i didn´t have to do step1.)

Thanks for the input. /Stefan

That is bizzare, what client / server versions are you running?

Server: 4.1.1.302
Client: SoftGrid for Terminal Servers Version: 4.1.0.56

We did an upgrade from an older client and server too, did not make any difference.

This was a problem with unloading of the softgrid/citrix users roaming profiles.
Not only did we get "Launch failed" but the system resources seemed to run out because of the roaming profiles not closing/unloadning properly.
The configuration changes in the SG client made it go away.

/Stefan

My primary site has this issue. I'm going to try the solution above and see.

Had a MS Premier engineer on site and he was an original Softricity guy. Here are tips we got from him regarding E005...

• The connectivity to the user settings folder may be suspect. (if they are on a network drive)
  
• The terminal server is overloaded. If your pool paged memory is over 80% allocated then this issue might happen. (look for 0k pkg files)
  
• The user settings file is corrupt.

Our issue is that the terminal servers were not configured to maximize the allocation of the pool paged memory so they were prematurely causing this error. You might want to toy with these settings but ultimately if you are using more than 80% of the pool then this issue will crop up.

HKEY_LOCAL_MACHINE\System\CurrentControlSet\Control\Session Manager\Memory Management

Value name: PagedPoolSize
Data type: REG_DWORD
Radix: Hex
Value data: 0xFFFFFFFF

Value name: PoolUsageMaximum
Data type: REG_DWORD
Radix: Decimal
Value data: 60
_________________