NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
This also sounds like you're trying to process a LOAD of stuff via pqact. I suspect you're getting bogged down with io requests.
The link you've found is a good one, and the -o 1 trick is one I've used in the past when I was trying to restart after problems.
gerry On 03/20/2012 03:36 PM, Adam DeConinck wrote:
Thanks all for the quick replies! We seem to be recovered for the moment, and will be revisiting our queue configuration. We may also be imposing limits using ulimit after a quick review, but I was trying to determine if LDM had any built-in throttling mechanism available. I always prefer to let the application limit itself if possible before letting the OS beat on it. We also found the following thread that has potentially useful things to say about this type of issue: (http://www.unidata.ucar.edu/support/help/MailArchives/idd/msg05853.html) I will have to determine from the users of the data whether limiting the amount of the backlog to download, with -o, is acceptable from their perspective. Thanks again, Adam On Tue, Mar 20, 2012 at 3:24 PM, Patrick L. Francis <wxprofessor@xxxxxxxxx> wrote:reduce the size of your product queue, delete and rebuild your queue, then adjust ldmd.conf to only one feed such as IDS-DDSPLUS to diagnose issues... setting limits should not be needed, but if you feel strongly about it you might want to google 'ulimit' cheers, --patrickHi all, We're currently attempting to diagnose a problem on an LDM node which we believe has a major backlog of data to download. When we attempt to start LDM, it quickly takes over all available memory and the load skyrockets, until the server chokes and dies. Server has 8 cores and 32 GB memory, and 8 GB swap. As of right now, the load is over 600 and nearly all memory is in use. We are not sure exactly what the cause of this issue is in LDM... only that as soon as we start LDM, the load goes through the roof. This issue occurred once before, which we solved with a memory upgrade 8GB->32 GB. Now it's happening again with more memory. :P LDM version is 6.8.1. Can anyone suggest a way to instruct LDM to rate-limit itself, or otherwise set a resource limit so it doesn't murder the server it's on? Thanks, Adam
ldm-users archives: