The issue we're fighting on our single system is a very large number of
connections. We normally have well over 1000 ldmd child processes
running. It takes ~1 hour to restart our LDM, due to the number of
connections it must re-establish. We're hoping that spreading this out
over a cluster will allow us to enable changes with no loss of data flow,
while also spreading that load out.
Matt
On Tue, Apr 5, 2016 at 1:43 PM, Chuck Pavloski <cfp102@xxxxxxxxx> wrote:
> Art Person has this running at Penn State.
>
> - Chuck
>
> On Apr 5, 2016, at 2:26 PM, Matthew Foster - NOAA Federal <
> matthew.foster@xxxxxxxx> wrote:
>
> We're looking into standing up an LDM cluster on virtualized servers, and
> are wondering if anyone else has done this. We are wanting to learn of any
> experiences or pitfalls that might help us along this path. We would
> likely set up something very similar to the cluster described on this page
> <http://www.unidata.ucar.edu/newsletter/2005june/clusterpiece.htm>, but
> in VMs, rather than bare-metal servers.
>
> Thanks for any suggestions or advice.
>
> Matt
>
>
> --
> Matt Foster
> Techniques Development Meteorologist
> NWS Central Region Headquarters, Scientific Services Division / STI
> Kansas City, MO
> Office: 816-268-3113
>
> _______________________________________________
> ldm-users mailing list
> ldm-users@xxxxxxxxxxxxxxxx
> For list information or to unsubscribe, visit:
> http://www.unidata.ucar.edu/mailing_lists/
>
>
>
> --------
>
> Chuck Pavloski, Ph.D.
> Director of Meteorological Computing and Senior Research Associate
>
> Department of Meteorology
> The Pennsylvania State University
> 503 Walker Building
> University Park, PA 16802
> USA
>
>
>
>
>
--
Matt Foster
Techniques Development Meteorologist
NWS Central Region Headquarters, Scientific Services Division / STI
Kansas City, MO
Office: 816-268-3113