Client and/or Designer Sessions Disappear

I had some anomalies trying to while run client and designer sessions last evening while working from home via a VPN connection.

My project is pretty simple with a header, left window with tabstrip which swaps out several pages have either historical easy chart trending or a table returned from a query. Each page is swapped out with a tab strip. Opening a trend page, the pens would load, and then the designer or client would freeze and after about 15-30 seconds would disappear. This was repeatable - I was able to reproduce multiple times. I also tried restarting the vision module, the server and the client machine.

At first I thought it query speed. I increased poll rate, but then I read the description that it’s not used for historical trending. Opening the pages with a query being returned to a table didn’t cause the issue. Turning off gateway communication in the designer and navigating around also seemed fine.

When the issues occurred, I didn’t see any errors that seemed relevant. I have some reoccurring opcua endpoint error and some driver timeouts, but these are still occurring today and all seems well here locally (no VPN). Also, looking at which the trends initially populated with data over VPN and locally the speeds were just about identical, maybe about a half second slower over VPN.

Looking at the gateway status after the disappearance it seems the gateway still thought the client existed. In fact, I was able to start another designer, connect and get a lock notice for the window that had been open in the previous designer session. Eventually, the sessions would go away in the gateway monitor, and there is a wrapper log note about closing timed out sessions.

Any clues to what may have been going on?

Sounds like a JVM hard crash. Look for JVM dump files on your hard drive - they’ll start with “hs_err_pid”, and might be located at C:\ or in your user dir or in Java’s dir or in the Windows dir. You might want to try upgrading the version of Java you’re running.

You are correct, I do have “hs_err_pid” files - one for each crash. I upgraded to Java 1.6.0_20 from 1.6_19. The server is running 1.6_17, but I’m guessing thats not part of the problem. I had the same results; crash over VPN , fine locally.

I’m confused why the crashes would only happen using a VPN connection. I guess I’ll try doing a complete java uninstall / reinstall to see if that helps. Do you have any other suggestions?

Post one of the crash files here (they’re just text files), it might give a clue.

It may not have anything to do with the VPN, but rather the computer you’re VPNing from.

I’m not sure my implication that it is same machine in both cases is working, sorry I should be more explicit. It’s a laptop in both locations. The only difference is VPN. I can browse the trend pages fine all day locally, no crash. Within a minute of opening one remotely the crashes occur.

I’ve attached a couple files. They all seem slightly different to me. Thanks.
hs_err_pid6948.log (18 KB)
hs_err_pid1312.log (16.9 KB)
hs_err_pid6328.log (14.4 KB)

I uninstalled and reinstalled java and it looks like I still have the same issue. I thought it was all better as my client stayed connected longer than I had been experiencing, but I believe it was how I was using the client.

What I noticed was if I’m on a trending screen and don’t navigate away to another screen until the Loading message underneath the bottom right side of the trend goes away, all is well. If the date/time bar isn’t populated and I try and navigate away, the crash occurs.

I was able to copy some output from the Java Console before it disappeared, and I’m attaching to see if it might shed some light on what is going on. It is showing an ArrayIndexOutOfBounds.
2010-06-7 java console output.txt (7.48 KB)

So it turns out it must have been my VPN client in conjunction with some new hardware that I didn’t know about :unamused:

We have multiple VPN entry points to our wide area network. I usually use our PA entry point since that is location I am based (and where the Ignition server resides in this case). In troubleshooting this problem, I decided to try our MI entry point. When connecting via this location, everything worked flawlessly.

Based on this, I immediately started to query my IT buddies. Well, it turns out they just replaced the VPN solution in PA from a Cisco Concentrator/Pix solution to a Cisco ASA solution. Of course, they didn’t inform anyone they had done so :angry:. I guess all the existing VPN clients that were deployed were supposed to work with the new hardware with no discernable changes. Actually, with the exception of the Java crashes when using Ignition that I have been experiencing and describing here, that seemed to be the case.

I upgraded to the Cisco AnyConnect VPN client, and what do you know – no more problems. I actually confirmed by switching to the older client which promptly caused the crash to occur.

Wow, thats pretty wild. I’m glad you figured it out.