This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.
Gini, > Please see Steve's reply below. I think Unidata believes that you can > more effectively present these issues without an intermediary, but I am > glad to assist in any way and appreciate being cc'd on your correspondence. > > I do see you have configured tgdv49 for upstream LDM pqact > [galvinvg@tgdv49 etc]$ grep pqact ldmd.conf > EXEC "pqact" > > while tgdv48 has > #EXEC "pqact" > [galvinvg@tgdv48 etc]$ > > And you said you are feeding the queue every 6 seconds: > > 18 files inserted as a single batch (one call to pqinsert) > into the upstream LDM product queue at 21:54:04 UTC (actually about 21:54:01 > according to the downstream LDM receiver's system clock). There were > subsequently eight more such batches inserted every five or six seconds (the > way we actually received them from the NCF) into the upstream product queue > (the last being at 21:54:31) before they all began to appear on the > downstream LDM system local disk filestore--totaling 333 Level III Radar > Products. The time-stamp on them all was 21:54:28 set according to the > > > But Wilson wonders if according to > http://www.unidata.ucar.edu/software/ldm/ldm-current/basics/behavior.html#downstream > the upstream LDM may be going to sleep: > > > 1. > > If the process reaches the end of the product-queue > <glindex.html#product-queue>, then it suspends itself for 30 > seconds or until reception of a SIGCONT <glindex.html#SIGCONT> > signal (from an ingester <glindex.html#ingester> or downstream LDM > <glindex.html#downstream%20LDM>) whichever occurs first. If the "pqinsert" process is being run outside the LDM system (i.e., it isn't being run due to an EXEC entry in the LDM configuration-file, etc/ldmd.conf) then the upstream LDM process on the upstream host will not receive any CONT signals from the "pqinsert" process and, consequently, will have to rely on its failsafe behavior of waking-up every 30 seconds to check for new data-products in the product-queue. The solution is to run the "pqinsert" process (indeed, all LDM ingester processes) out of the LDM configuration-file. Regards, Steve Emmerson Ticket Details =================== Ticket ID: OAO-873149 Department: Support LDM Priority: Normal Status: Closed