LAPS Users Discussion Forum - Problems with 43-12

Matt Foster

LAPS User

Offline

Rock Chalk Jayhawks!

Posts: 26

Problems with 43-12
Apr 8^th, 2011 at 1:59pm

I have two problems (that I've found so far) with 43-12 that I installed earlier this week. For some reason, profilers and cloud drift winds have stopped ingesting.

ingest_cloud_drift gives the following error every time it executes...

PGFIO-F-231/formatted read/internal file/error on data conversion.
In source file wfo_fname_conversions.f, at line number 19

As for ingest_pro, it isn't as clear what's going wrong. The log indicates that it finds the input file (AWIPS NetCDF), but then doesn't seem to read the data correctly. For example, at 13Z this morning there are 27 stations in the source file, but ingest_pro says it only finds 2. It goes on through the BLP routine, which we don't use, then to the VAD (NIMBUS) ingest, and it stops. The last message in the log is: "Return from VAD (NIMBUS) ingest", and no .pro file is written.

The cloud drift winds aren't a big deal to us, but the profilers are.

Help?

Oh...this is with the pre-compiled, static 64-bit binaries.

Matt @ OUN

Back to top

IP Logged

Steve Albers

Administrator

Offline

I Love YaBB 2.4!

Posts: 48
Boulder, CO

Re: Problems with 43-12
Reply #1 - Apr 8^th, 2011 at 10:33pm

With cloud drift - are there any files in the input data directory that are extraneous and don't have the regular time stamp in the file name? Apparently the numerical string for the time doesn't have numbers in it anymore for some reason, perhaps internally in the code.

Otherwise, it might be interesting if you have the older version still to see if that is working better on these datasets.

We may have an AWIPS machine here that has the same type of data if we can test some of these newer LAPS versions on it.

Back to top

IP Logged

Matt Foster

LAPS User

Offline

Rock Chalk Jayhawks!

Posts: 26

Re: Problems with 43-12
Reply #2 - Apr 11^th, 2011 at 12:11pm

Steve,

You hit the nail on the head with the cloud-drift winds. We're getting those from MADIS, and they weren't getting gunzip'd. I'm working that issue now.

As for the profilers...it seems very odd. We've had 4 runs in the last couple of days, where the profilers decoded correctly. On all of the other runs, it reads in the first two stations from the file, which are Alaskan profilers, ingest_pro says they are both out of the area, and it returns from WPDN ingest without reading any other stations. AFAIK, it was working fine on 41-12, which was the previous version we were running.

Matt

Back to top

IP Logged

Matt Foster

LAPS User

Offline

Rock Chalk Jayhawks!

Posts: 26

Re: Problems with 43-12
Reply #3 - Apr 12^th, 2011 at 1:38pm

Steve,

The profiler issue seems to be a case of ingest_pro running before the data has become available. I kicked off sched.pl at :21, instead of :20, and the profilers decoded OK. It appears that everything leading up to the execution of ingest_pro got slightly faster between 41-12 and 43-12, such that ingest_pro is executing almost a minute earlier. Since the profiler data isn't available to us until between :20 and :21, that was just enough to throw a wrench in the works.

Matt

Back to top

IP Logged

Steve Albers

Administrator

Offline

I Love YaBB 2.4!

Posts: 48
Boulder, CO

Re: Problems with 43-12
Reply #4 - Apr 12^th, 2011 at 5:24pm

Glad to hear this is working better now, on both datasets.

The software such as LGA is more efficient now, so this may be a factor for the profiler. The profiler ingest does have a wait function that with hourly data should wait for the file to show up. If there are only a few profilers though it might pass the wait test, since the file at least is there.

Back to top

IP Logged

Matt Foster

LAPS User

Offline

Rock Chalk Jayhawks!

Posts: 26

Re: Problems with 43-12
Reply #5 - Apr 13^th, 2011 at 1:01pm

Steve,

Since I changed sched to run at :21, instead of :20, profilers have decoded on every run. Previously, it was finding the file every time, but it only had the 2 Alaskan sites. It isn't clear at all what was happening. I could run ingest_pro from the command line shortly after it failed when run by sched, and all sites would decode normally.

Of course, this introduced a new problem. LAPS is only barely finishing in time for the start of our model run at :30, so I either need to get LAPS faster or push back the start of the model by a few minutes.

Matt

Back to top

IP Logged

Steve Albers Administrator Offline I Love YaBB 2.4! Posts: 48 Boulder, CO	Re: Problems with 43-12 Reply #6 - Apr 14^th, 2011 at 6:15pm I wonder if the 2 Alaskan profilers are coming in slightly ahead of the others in being written to the file?
Back to top	IP Logged

Matt Foster LAPS User Offline Rock Chalk Jayhawks! Posts: 26	Re: Problems with 43-12 Reply #7 - Apr 15^th, 2011 at 12:03pm Steve, Yes, they do. I checked the profiler log on AWIPS, and the two Alaskan sites come in around :15, and all of the others come in between :20 and :21. Matt
Back to top	IP Logged

Steve Albers Administrator Offline I Love YaBB 2.4! Posts: 48 Boulder, CO	Re: Problems with 43-12 Reply #8 - Apr 18^th, 2011 at 9:14pm We hope to send out a new release this week that will have a superobing feature for temp.exe and wind.exe. The parameter for this will be in 'nest7grid.parms'. Maybe this will help the runtimes a bit? Other programs are a bit more efficient as well. Steve
Back to top	IP Logged

Matt Foster LAPS User Offline Rock Chalk Jayhawks! Posts: 26	Re: Problems with 43-12 Reply #9 - Apr 20^th, 2011 at 2:30pm Steve, qbalpe is taking a couple of minutes to complete also. Any improvements there? Matt
Back to top	IP Logged

Steve Albers Administrator Offline I Love YaBB 2.4! Posts: 48 Boulder, CO	Re: Problems with 43-12 Reply #10 - Apr 20^th, 2011 at 6:06pm Matt - I made a change to some inner do loops in 'leibp3' for the coming release in the balance code. I'm in the process of evaluating how much speedup there is. Steve
Back to top	IP Logged