VMware Horizon Community
surferdave
Enthusiast
Enthusiast

View clients randomly disconnecting

I have searched around, but I haven't found anything on this issue already posted.  Please forgive me if I overlooked an existing article.

We have 40 clients running 4.5 and Windows XP.  The users will get their connection dropped and have to log back into the connection server.  The session is persistant so they don't lose work, but they are becoming more and more frustrated with dropped connections.

I am currently investigating our Cisco switches to make sure they are all up to date.  I receive the following in the event log of the client VMs, and searches haven't revealed anything helpful.  The events coorilate to the times connections are being dropped:

Event Type:    Information
Event Source:    VMware View
Event Category:    VMware View
Event ID:    102
Date:        2/21/2011
Time:        2:02:24 PM
User:        NT AUTHORITY\SYSTEM
Computer:    IT-THIN-1
Description:
Needed to disconnect existing PCoIP connection during requestConnection

Event Type:    Information
Event Source:    VMware View
Event Category:    VMware View
Event ID:    102
Date:        2/21/2011
Time:        2:02:24 PM
User:        NT AUTHORITY\SYSTEM
Computer:    IT-THIN-1
Description:
PCoIP requested disconnect while busy!

Event Type:    Information
Event Source:    VMware View
Event Category:    VMware View
Event ID:    102
Date:        2/21/2011
Time:        2:02:24 PM
User:        NT AUTHORITY\SYSTEM
Computer:    IT-THIN-1
Description:
Closed PCoIP connection doesn't match global value

Event Type:    Information
Event Source:    VMware View
Event Category:    VMware View
Event ID:    102
Date:        2/21/2011
Time:        2:02:25 PM
User:        username
Computer:    IT-THIN-1
Description:
Unable to locate route for response queue [[51192fe3-2c96-49e9-acc3-2f0ed484ed01]MessageFrameWork_1.

If anyone has seen this issue, please let me know.

Thanks,

Dave

Reply
0 Kudos
18 Replies
mittim12
Immortal
Immortal

I assume your running PCOIP since all of the errors reference it in some way.   Did you look over the PCOIP logs to see if you can get any additonal information?   

Do the users all get dropped at the same time?    PCOIP uses a direct connect model so it shouldn't  have any dependecy on the connection brokers other than to determine which desktop to login to. 

Reply
0 Kudos
surferdave
Enthusiast
Enthusiast

I looked over the log files, nothing stands out.  Is there something specific I should look for?

Reply
0 Kudos
surferdave
Enthusiast
Enthusiast

Thanks, we increased the timeout to max a few weeks ago.  We are running ESX 4.1.0 and View 4.5.  I am not sure where you would check the SSL for the clients, looks like the location is moved in View 4.5.

Reply
0 Kudos
alefestaedist
Hot Shot
Hot Shot

Under Global Settings..

Attached a picture to help you. 

Reply
0 Kudos
surferdave
Enthusiast
Enthusiast

Okay, I will try disabling the SSL.  I thought that setting was anyone connecting to the Administrator server (via HTTPS), not the clients.

Thank you.

Reply
0 Kudos
alefestaedist
Hot Shot
Hot Shot

You well anywa let me know if it worked for you.

Alex

Reply
0 Kudos
mittim12
Immortal
Immortal

Nothing in particular.  Was just wondering if there was more information that what was available in the event viewer.     If everyone is being disconnected at one time I would say it's a network issue.  if it's kind of random as your title suggest then a good place to start may be the KB that the other user specificed.

I do like the KB resolution.   Just disable SSL for clients and that's it.   Nothing about what is actually causing the issue Smiley Happy 

Reply
0 Kudos
alefestaedist
Hot Shot
Hot Shot

Found this in the "installation guide of vmware view 4.5"  and wonder if couldnt' be this:

"Verify that the View Client setting for using secure (SSL) connections matches the global setting in View

Administrator. For example, if the check box for secure connections is deselected on the client, the check

box must also be deselected in View Administrator."

Reply
0 Kudos
mpryor
Commander
Commander

I think it's best to backtrack slightly. Firstly the messages such as "Needed to disconnect existing PCoIP connection during requestConnection" are a symptom of the previous disconnect during the user's reconnection attempt - they suggest that the agent hasn't been told the previous pcoip connection had been dropped, and needed to be cleaned up.

Secondly, I just took a look at KB 1023979 and it's not likely to be related, as a requested disconnect would clean up the PCoIP server process cleanly - you wouldn't see the errors in the event log you do. The SSL setting certainly would make no difference even if it was. (the KB is not particularly clear, it is meant to be talking about the specific case of the broker requesting a desktop session is disconnected when the broker logon expires, but is missing the key line in the agent log that says "received a DisconnectSession message" - I've contacted the author).

You may not have spotted anything obvious from the pcoip_server logs but I think you should attach them anyway, if the connection is dropping that's where the hint for the root cause will be.

Mike

Reply
0 Kudos
alefestaedist
Hot Shot
Hot Shot

Mike,  I agree that the whole pcoip_server logs file would help as that the messages like Needed to disconnect existing PCoIP connection during requestConnection" are related to a previous dropped connection, but I focused in reply to the message "Closed PCoIP connection doesn't match global value" that for the few information that we have could be a good point to start from, my fault to not ask for logs instead to give "useless" answers.

Alex

Reply
0 Kudos
mpryor
Commander
Commander

Hi Alex, I certainly didn't mean to imply your answer was useless! I found the KB pretty unclear and too generic when I read it, which isn't a good sign because I knew what it was trying to descibe in advance. The error logs pasted at the top of this thread could be from any scenario where a (re)connection request comes in when the agent has stale information on the last pcoip session, and the logs in the KB are not all directly relevant to the issue it is trying to describe. As I said, I'm going to follow up to get the KB improved. Just trying to save you guys from going down the wrong investigation path.

Reply
0 Kudos
alefestaedist
Hot Shot
Hot Shot

Mike, I tell myself was useless LOL..what I forgot is the first rull of troublshooting "start from the log" ..is a lesson learned, always ask for the complete log and than start the investigation even in a community.

Anyway I hope to see the log I'm definetly curios to see what's the problem and how to solve it.

Alex

Reply
0 Kudos
surferdave
Enthusiast
Enthusiast

I changed the SSL settings yesterday and I will test today to see the results.  I also upgraded our network switches based on another article just to make sure we are not having a switch issue.

The clients do not all drop at the same time, they are random times for each client.

NOTE: If you change the SSL setting, make sure the clients are all setup to not use SSL.  We are using the local group policy ADM files on the client machines to specify the server.  Unless you put a http://servername in the policy, the client will try and connect using SSL and fail. (found this out the hard way).

Reply
0 Kudos
jhilbish
Contributor
Contributor

did removing the ssl selection fix the issue?  i am seeing the exact issue

Reply
0 Kudos
jburleigh
Contributor
Contributor

We were experiencing the same issue in our environment that was running View 4.5 and ESX 4.0.   Since then we have upgraded to View 4.6 and ESXi 4.1 and I have not seen the issue.

I have increased the RAM on the View connection server to 8GB for about 40 clients running Windowx XP SP3 and Windows 7 SP1.

We are running Wyse P20 devices and I have upgraded the firmware to 3.3.

My theory is that the random disconnects are related to bandwidth used and or memory utilzation of the host.  I have notice at times when viewing Youtube or have multiple applications open that the disconnect would happen. 

VMs are configured with 30GB of disk, 3 GB of Ram and 1 CPU.

Backend infrastructure is IBM x3650m3 with 48GBb of RAM in a clustered solution.  Backend storage is EMC CX4 with clients divided on two 800GB LUNS.

Since the upgrade I have not experience one disconnect and UI performance has increased.

John

Reply
0 Kudos
RajuVCP
Hot Shot
Hot Shot

Earlier i faced the same problem. later found its a Network issue. packets droping causes it. now it resolved. Smiley Happy

Regards

Raju Gunnal

http://www.tech2solution.com/forum

Raju Gunnal VCP 4, VCP 5, VTSP 4, VTSP 5, ITIL V3 http://www.techtosolution.com
Reply
0 Kudos
VRudi
Contributor
Contributor

I had the same issue, just with view 4.5 and esx 4.0

eventually it was an issue on one of the hosts, I tracked the disconnected users and they all came from the same host. Reloaded the host and everything was fine

Reply
0 Kudos