Their sessions last anywhere between a half hour and hours. We're using DNS Round Robin which I know is probably the culprit. But I didn't get this much feedback while on Juniper SA. Now we're on the MAGs.
I tried changing the TTL to 8 hours as I noticed the user was logged into 1 MAG and the Pulse traffic in the client events were pointing to another MAG. I was almost certain this would resolve the issue but they are still reporting disconnects.
The common complaint is Citrix Receiver sessions close on them. Whether they do it through the web without pulse or launching Receiver after Pulse is connected. They end up losing their Receiver sessions. Pulse in most cases stays connected. I noticed switching wireless networks that Pulse just reconnects without any pop ups.
There doesn't appear to be any user access logs that tell me why the user got disconnected. However, in the events log it shows the channel to the mag had an error.
As I mentioned, these issues did not appear to be an issue on the Juniper SAs. I could be troubleshooting the wrong things here. Suggestions, comments?
We are experiencing exactly the same issue, users connect on fine and then RDP or citrix reciever drops connection but pulse still shows as connected. The user has to disconnect pulse and then reconnect to be able to reconnect to RDP or citrix, this is not consistent and can be fine for hours and then other times it can disconnect 3 or 4 times a hour.
Did you manage to fix your issue?
We are using a SA-2500 with firmware 8.0R7.1.
We had the issue with some users in our network. Two solutions were implemented.
First, we tried the Secure Pulse v5.1.1, which solved the issue for some of the users.
Second, we found out that a specific router caused the issue. A lot of the users used the same router or 3 different models o the same brand: D-Link. The solutino was implemented on D-Link DIR-615, DIR655 and DIR650.
In the advanced tab of the router's web interface>firewall settings>NAT Endpoint Filtering:
Change the settings for both UDP and TCP Endipoint Filtering to : "Endpoint Independent "
the issue disappeared completely after this manipulation.