Opendeploy crashes with Unknown socket error

cute_guy

We are running a cron job that deploys thousands of files from the filesystem path to LiveSite Runtime servers and TeamSite workarea, there are 6 legs of deployments that are run in parallel.

 

The OD crashes throwing error as 

Failed to receive message from OpenDeploy Server
Reason: sockErr=0, Unknown socket error
Check that the OpenDeploy server on host localhost is listening on port 3434.
Failed to receive message from OpenDeploy Server
Reason: sockErr=0, Unknown socket error

 

When checking the OD micro logs the processing has abruptly stopped in the middle.

 

Opendeploy 7.1 and Linux OS

Adam Stoller

I'd work on narrowing down the problem, if possible; things like:

• Reduce the deployment to a single leg (I assume you're using 'leg' in terms of receivers  - yes?)
• Does the problem persist?
• If not, add another leg and re-try
• If you've added several legs before the problem crops up again, remove the first leg and try again
• If the problem still happens you've identified a leg that causes problems and can focus in on that.
• If the problem persists with just one leg, try using a different leg and repeat above; if necessary go through all 6 legs individually
• Try reducing the number of files being deployed, start with 100, then 500, then 1000, etc.
• If it works with a lower number but fails with a higher one, split the difference and try again, continue doing so until you can identify a number (or small range of numbers) that represents the threshold
• If you can do that, check memory usage on both the Base and Receiver sides during deployments as well as checking the diskspace available on the disks (both the target location and tmp directory partitions)
• How large is your odnodes.xml file on your Base server?
• There was a bug in 7.2.0 that could easily have been in 7.1.0 as well - where the odnodes file was re-read everytime a deployment was run but didn't reliably free up memory. I found this on a project where the odnodes.xml file was something like 2500 lines long (perhaps 25000, don't remember right now - but it was BIG) and we kept running into problems due to the memory leak.
• Anything else you can think of to try to narrow down the source of the problem is worth doing (and taking careful notes of everything you do)