On Tue, 2 Aug 2011, Chad Johnson wrote:
All,
We found that the high rate of retransmitted products ended on Monday 08/01 - around
14:44 UTC. Up until that point, we were seeing retransmissions at a rate of thousands per
hour and the difference between the time of original transmission and the time of
retransmission was around 6000 seconds or between 1.5 to 2.0 hours. Then, there was a
half hour gap between 14:44 and 15:13 where no retransmissions were sent, followed by the
typical random low rate of restrans we're used to seeing. The time difference between the
original and the retrans product also settled down to a typical 30 and 60 seconds. After
8/01 14:44, I'd guesstimate that we had already seen 95% of the retransmissions. Before
that, I'd say it was the inverse (possibly because products were no longer in the
"seen it" list). Anyway...something happened at 14:44 on the 1st. Is anybody
aware of any changes made by the NCF/TOC around that time?
I talked to the TOC late yesterday afternoon and they said everything was fine.
Which I guess, technically, at the time was true but they didn't provide any
additional information.
What seems really odd to me, is the length of time between the original
transmission and the restransmission. It's almost as if there is a rogue NWS
office out there requesting retransmissions long after the product has been
transmitted. Here, a slug of radar products were retransmitted nearly two hours
after the original was broadcast. We were also seeing the same for some GRIB
data.
2011/08/01 00:03:07 Retranmission already exists: skipping "SDUS54 KOUN
312215"; Secs=6370
2011/08/01 00:03:07 Retranmission already exists: skipping "SDUS54 KOUN
312215"; Secs=6370
2011/08/01 00:03:07 Retranmission already exists: skipping "SDUS23 KIND
312211"; Secs=6370
2011/08/01 00:03:07 Retranmission already exists: skipping "SDUS23 KARX
312214"; Secs=6370
2011/08/01 00:03:07 Retranmission already exists: skipping "SDUS23 KIND
312211"; Secs=6370
They may have even retransmitted the same product multiple times, unless
multiple unique products are transmitted under the SDUS54 KOUN header (not up
on my WMO headers like I used to be).
2011/07/31 22:16:21 Retranmission doesn't exist: saving "SDUS54 KOUN 312215"
2011/08/01 00:01:34 Retranmission already exists: skipping "SDUS54 KOUN
312215"; Secs=6335
2011/08/01 00:01:34 Retranmission already exists: skipping "SDUS54 KOUN
312215"; Secs=6335
2011/08/01 00:03:07 Retranmission already exists: skipping "SDUS54 KOUN
312215"; Secs=6370
2011/08/01 00:03:07 Retranmission already exists: skipping "SDUS54 KOUN
312215"; Secs=6370
At that time, we lost all METARs. The NWS sent out a memo saying that a
SAN had gone crazy; they had switched to a "backup system" and that SAN
was having issues. Once fixed, the METARs came back and the duplicate
products stopped. So it wasn't the satellite broadcast or NWS offices
having trouble getting the broadcast.
*******************************************************************************
Gilbert Sebenste ********
(My opinions only!) ******
Staff Meteorologist, Northern Illinois University ****
E-mail: sebenste@xxxxxxxxxxxxxxxxxxxxx ***
web: http://weather.admin.niu.edu **
Twitter: http://www.twitter.com/NIU_Weather **
Facebook: http://www.facebook.com/niu.weather *
*******************************************************************************
