Hi,
I have several Relativity Servers deployed and recently some of the are presenting stability problems, even thought the service appears running on the Windows Services Manager, no client can reach it. If i stop and re-start the service, then it starts working again.
This issue has occurred with several versions of Relativity 7 and 8. Mostly on Windows Server 2012.
Any ideas?
Could you be running out of physical sockets on the machine, on OS level? i believe “netstat” will provide infos about open sockets.
Essentially what happens is that after connection closes, the socket goes into a mode where it’s not available for a while. and Windows has only a limited (65535) number of these guys.
Thanks for the reply, will check that and keep posting.
Hello, we have the same situation with some of our clients. Right now I have an installation with Windows Server 2008 R2 Service Pack 1 (Windows SBS Server 2011). We installed Relativity Server 8.0.83. This situation is happening at least once a day. The users say they get a timeout message, but the Relativity Server service shows up as Started. If we Stop and Start the service it works again. On this same installation we were using a different server and it did not have that problem.
We are monitoring the Relativity Server. Right now it is not responding and we noticed is that there are a lot of sockets with a CLOSE_WAIT status. Apparently this connections stay there and Relativity Server do not close them ?
CLOSE_WAIT is what i described above. I believe it means that the socket was closed correctly, but the OS puts it on hold for a given amount of time. But we will investigate this further.
thanx,
marc
Btw which channel type is used on the client side? Also on the server side do you use the default channel on 7099 or custom channel?
We are using the default port 7099. One particular issue is that this server is using Windows Small Business Server 2011 Standard. Do you have any issues with Relativity Server specifically with this server product?
So I’ll assume that on the client side one of the plain Http client channels is used.
Then it seems that on certain OS + .NET Framework combinations sockets aren’t properly freed when SuperHttp channel serves Plain-HTTP requests. This bug will be logged, however I cannot provide exact fix date for it as it would require a good amount of research.
As a workaround configure a second channel as Http on port 7100 and use it instead.
Thanks, logged as bugs://69749: Serious stability issues with Relativity Server
Anton,
Thanks for the information. to answer your question on the server side we are using Relativity Server v8.0.83.1137. On the (Network Settings ) Server Channel type is Http, port 7099, session manager type: InprocessSessionManager.
On the client, which is an XCode application we use ROHTTPCLientChannel.
Let me know if you need more information. Thanks for your support!
I have to ask again:
This installation is on a Windows Small Business Server 2011 Standard.
My customer is asking if Relativity is certified to run on this server version.
Do you have a list of Windows Servers versions that are certified or validated for Relativity Server ?
Relativity Server should run on all versions of Windows, yes. I do not expect this problem to be related to the specific Windows version. Our team is investigating this problem further.
Thanks mh,
As a follow up to this issue: yesterday we installed Relativity Server v8.1.85.1143 on the server of the customer in the hopes that this new version will clear the issue. However the issue has already happenned twice today: The service shows up as running but it does not respond, not even if we try to call it by typing the http://address:7099 from a browser.
I am including a report of the TCP connections status In the moment the service is not responding. Maybe you can make some sense of it.
The solution is to stop and start the Relativity service. We have done so twice today already.
informe.txt (2.1 MB)
Will appreciate any insight with this situation.
I apologize for that.
8.1.85.1143 is indeed not supposed to have a fix for this yet, as we’re still investigating, so in a way thats good/expected. I will make sure to chase this hop with the team to see where we are, on monday. I realize this s urgent for you, and i consider this a critical issue we need to address soon.
thanx for your patience,
marc
I think we might have a fix for this, finally. We’re shooting for a new beta build coming early next week.
Was this ever fixed? I’m having the exact same problem on a Windows Server 2008 R2 machine.
Everything runs flawlessly until suddenly it’ll stop accepting remote requests… The service will keep running and can be accessed locally, therefore no errors are ever logged to the event viewer.
If the relativity.exe application is restarted service will resume accepting remote connections.
What is the exact server version / channel type you use?
Simple Http Channel
Relativity Server: v9.0.97.1245
Data Abstract: v9.0.97.1245
RemObjects SDK: v9.0.97.1245
Internet Pack: v2.0.97.1245
Framework: v4.0.30319.36365
Operating System: Microsoft Windows NT 6.1.7601 Service Pack 1
Platform: Win32NT
Just to confirm - you work via this channel, not via default one at port 7099 ?