LAE

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: johnvvu

What are some ways to see if the node is still running? Some of our users dock and undock their laptops, which causes a drop in the network connection. I noticed that the nodes still shows a yellow cogs wheel as if the node is stilling running. However, when I run a ls -l on the tmp file associated with node, I see that the file was growing in size and leads me to believe the node is not running. Is this a reliable way to see if the node is still running?

Please advise.

Thank you

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: ltolleson

John,

You will get varying degrees of behavior when BRE is disconnected from the server while a node is running. There is a sequence of events that occur when a node is executed in BRE and breaking the connection disrupts that sequence of events. First, BRE communicates with the LAE server to start the node execution. Next, nodes that have been started will continue to run until completion at which time LAE will communicate back to the BRE session which started the node and BRE will update the status of the node based on information sent back from LAE. The problem is when the network connection between BRE and the server occurs while a node it running. The server does not know how to re-connect to BRE, so the process will become orphaned and continue to try to communicate with BRE until it times out.

If you check the temp brd file on the server for a node and the file size continues to grow, then that process is still running and should run to completion. The problem is that even if you reconnect and the node completes, the node status will not update since LAE cannot re-connect back to BRE.

There are a few ways that you can check for orphaned nodes. The first is what you described. Checking the size of the temp file will let you know if the node is still writing output. Another way is to look at the system processes. You can use the command "ps -el |grep lae". This will show you the processes that are running under the lae username. If there are any LAE nodes running you should see a process related to that node. If you suspect that a node is orphaned and is running for a long time you should have your system admin either kill the node or re-start the LAE server.

If your users have to un-dock often I would recommend using a wireless network, so that network connectivity is not lost when the un-dock.

We have also experienced this same issue when a user is running a graph over a VPN connection and the connection to the VPN is lost. This scenario also causes issues when nodes are running.

Hope that helps

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: johnvvu

Thank you for the information-that's helpful. I've been checking the tmp files but will also use the ps command. You mentioned Killing the node, is that something I can do with the execute system command? Or Is there a permissions issue here?