[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: compare desi to tg2 (fwd)
- Subject: Re: compare desi to tg2 (fwd)
- Date: Tue, 20 Jul 1999 16:00:58 -0600 (MDT)
===============================================================================
Robb Kambic Unidata Program Center
Software Engineer III Univ. Corp for Atmospheric Research
address@hidden WWW: http://www.unidata.ucar.edu/
===============================================================================
---------- Forwarded message ----------
Date: Tue, 22 Jun 1999 11:31:22 -0500
From: Chad Johnson <address@hidden>
To: Robb Kambic <address@hidden>, address@hidden
Subject: Re: compare desi to tg2 (fwd)
Robb Kambic wrote:
>
> Chad,
>
> I created a ldm release ldm-5.0.8 that appends the PIL to the WMO header
> so users can key on it. It's a modification to pqing, etc. Can you
> install this release on you noaaport machine? I think Chiz wants to
> announce the PIL today.
OK. I'll get it installed today. I'll let you know when I'm finished.
> Your ideas about comparing the 'bit error' messages is good one, but the
> tg2 machine is no longer connected to the NOAAport feed. The tg2 machine
> is the one that is getting prepared to ship to LSU. If the situation
> arises again then I'll check the messages.
We think we found the culprit to our bit error problem and I wonder if it might
be related to the problems you are seeing. It appears as though there is a bug
in
SolarisX86 2.5.1 when memory mapping files in a tmpfs file system which have
been
modified, but not removed (which is what happens with the latest mod of inge).
To
fix the problem we moved the creation of the spool files from /tmp (a tmpfs file
system) to /data (a ufs file system). Please check which OS version you are
running on both your systems. We don't appear to be having this problem with
SolarisX86 2.6 or 7. If you are running 2.5.1 with the latest inge mods and the
spool files are being placed in /tmp, we recommend that the spool files be
placed
in /data. This can be modified by editing the file /etc/ingcntl. Under the
'start' case, change the line which reads 'cd /tmp' to 'cd /data'. Stop and
restart the ingestor.
-Chad
>
> Thanks for installing the new ldm release,
> Robb...
>
> On Thu, 10 Jun 1999, Chad Johnson wrote:
>
> > Robb Kambic wrote:
> > >
> > > Chad,
> > >
> > > Here's some more info Chiz uncovered. Actually this seems like a timing
> > > error for inge, just a thought.
> >
> > The ingestor software writes error messages such as 'bit error', 'hdlc
> > problem',
> > 'overhang', etc when it encounters a problem with the datastream. If one of
> > these
> > problems occur when a product is in transmission, this product will be
> > dropped.
> > Could you compare these error messages between the two systems and see if
> > they
> > are identical? If any errors do not occur at the same time on both systems,
> > this
> > may account for the differences you are seeing in the products reaching the
> > LDM.
> >
> > -Chad
> >
> > >
> > > Robb,,,
> > >
> > > ===============================================================================
> > > Robb Kambic Unidata Program Center
> > > Software Engineer III Univ. Corp for Atmospheric
> > > Research
> > > address@hidden WWW: http://www.unidata.ucar.edu/
> > > ===============================================================================
> > >
> > > ---------- Forwarded message ----------
> > > Date: Wed, 9 Jun 1999 12:53:48 -0600
> > > From: Steve Chiswell <address@hidden>
> > > To: address@hidden, address@hidden
> > > Subject: compare desi to tg2
> > >
> > > This morning Mike reverted to the previous inge on tg2
> > > for comparison against desi running SSEC's latest version.
> > >
> > > Here are some comparisons:
> > >
> > > from 1640Z through 1822Z, the files /scratch/chiz/tg2.log1
> > > and /scratch/chiz/desi.log1 are sync'd to both start with
> > > product 738 and end with product 982 for DDPLUS|IDS products.
> > >
> > > %wc -l tg2.log1 desi.log1
> > > 7833 tg2.log1
> > > 7804 desi.log1
> > >
> > > tg2 obtained 29 more DDPLUS products.
> > >
> > > Comparing both files with just the WMO header parts shows that
> > > both machines saw some products the other one didn't. This
> > > would seem to indicate that the latest version does not fix the previous
> > > problem we were seeing with missing products.
> > >
> > > diff tg2.sav desi.sav
> > > 122a123,126
> > > > IDS|DDPLUS 214 FQUS23 KWBC 091638
> > > > IDS|DDPLUS 215 SIVD80 KWBC 091500 RRD
> > > > IDS|DDPLUS 216 SIVF30 KWBC 091500 RRB
> > > > IDS|DDPLUS 217 SPCN43 CWAO 091638
> > > 563,565d566
> > > < IDS|DDPLUS 891 WSCN33 CWEG 091646
> > > < IDS|DDPLUS 892 WWIN40 DEMS 091200
> > > < IDS|DDPLUS 893 WSCN34 CWEG 091647
> > > 1092d1092
> > > < IDS|DDPLUS 708 WWCN31 CWXE 091657
> > > 2179,2194d2178
> > > < IDS|DDPLUS 941 SAAK31 KWBC 091600 RRE /pMETAR
> > > < IDS|DDPLUS 942 SACA42 KWBC 091700 /pMETAR
> > > < IDS|DDPLUS 943 SACA42 KWBC 091600 RRA /pMETAR
> > > < IDS|DDPLUS 944 SAHW31 KWBC 091700 /pMETAR
> > > < IDS|DDPLUS 945 SAPA41 KWBC 091700 /pMETAR
> > > < IDS|DDPLUS 946 FEXW40 KWBC 091200
> > > < IDS|DDPLUS 947 FEXW41 KWBC 091200
> > > < IDS|DDPLUS 948 FEXW42 KWBC 091200
> > > < IDS|DDPLUS 949 FEXW43 KWBC 091200
> > > < IDS|DDPLUS 950 FEXW44 KWBC 091200
> > > < IDS|DDPLUS 951 FEXW45 KWBC 091200
> > > < IDS|DDPLUS 952 FEXW46 KWBC 091200
> > > < IDS|DDPLUS 953 FEXW47 KWBC 091200
> > > < IDS|DDPLUS 954 FEXW48 KWBC 091200
> > > < IDS|DDPLUS 955 FEXW49 KWBC 091200
> > > < IDS|DDPLUS 956 FEXW50 KWBC 091200
> > > 3240a3225,3226
> > > > IDS|DDPLUS 308 FTPA31 PHNL 091700
> > > > IDS|DDPLUS 309 UBUS01 KNKA 091722
> > > 3356,3367d3341
> > > < IDS|DDPLUS 489 UAPA01 KWBC 091722
> > > < IDS|DDPLUS 490 FXUS07 KBOI 091718 PAA /pFWFBOI
> > > < IDS|DDPLUS 491 SPCN43 CWAO 091720 CCA
> > > < IDS|DDPLUS 492 UACN10 CYBW 091722
> > > < IDS|DDPLUS 493 FTAS31 VTBB 091600 RRB
> > > < IDS|DDPLUS 504 SRUS53 KABR 091721 /pRR3ABR
> > > < IDS|DDPLUS 510 SRUS55 KMSO 091721 /pRR1MSO
> > > < IDS|DDPLUS 512 FTVN20 SVMI 091618
> > > < IDS|DDPLUS 513 UAXX36 KAWN 091700 RRW
> > > < IDS|DDPLUS 514 FPAK20 PANC 091722
> > > < IDS|DDPLUS 515 FACN33 CWUL 091730
> > > < IDS|DDPLUS 516 FPUS74 KCRP 091722 /pNOWCRP
> > > 5209a5184,5188
> > > > IDS|DDPLUS 842 CXUS33 KOAX 091752 /pSCDOAX
> > > > IDS|DDPLUS 843 SRUS76 KHNX 091750 /pRR7FAT
> > > > IDS|DDPLUS 845 SRUS73 KFSD 091750 /pRR7SUX
> > > > IDS|DDPLUS 848 SXUS01 KNYC 091753 /pOSONYC
> > > > IDS|DDPLUS 849 SXUS42 KTAE 091755 /pOSOTLH
> > > 5429d5407
> > > < IDS|DDPLUS 089 SXUS70 KWAL 091756
> > > 5913,5919d5890
> > > < IDS|DDPLUS 646 SRUS71 KILN 091800 /pRR7CVG
> > > < IDS|DDPLUS 647 SRUS71 KGYX 091800 /pRR7AKQ
> > > < IDS|DDPLUS 648 SRUS73 KLOT 091800 /pRR7P75
> > > < IDS|DDPLUS 649 SRUS76 KMTR 091800 /pRR7SCK
> > > < IDS|DDPLUS 651 SRUS73 KGRR 091800 /pRR7GRR
> > > < IDS|DDPLUS 652 SRUS72 KFFC 091800 /pRR7AHN
> > > < IDS|DDPLUS 653 SRUS72 KFFC 091800 /pRR7ATL
> > >
> > > files tg2.log2 and desi.log2 contain HDS products from
> > > 732 through 905 from 1639Z to 1830Z.
> > >
> > > %wc -l tg2.log2 desi.log2
> > > 4600 tg2.log2
> > > 4609 desi.log2
> > >
> > > desi obtained 9 more hds products.
> > >
> > > Diffing the WMO header portion of these files shows again that both
> > > ingesters missed different products. Note the difference between:
> > > < HDS 201 HUJK20 KWBC 091200 /m93
> > > ---
> > > > HDS 201 HUJK20 KWBC 091200 /mAVN
> > >
> > > This difference in the WMO header is cause by a difference in the bits
> > > that pqing uses to identify the model number in the PDS block of the grib
> > > product. This would rule out a satellite dish error since both machines
> > > see the same data stream feed. The difference in the products
> > > would therefore have to be originating in the spool files.
> > >
> > > diff tg2.savhds desi.savhds
> > > 513a514
> > > > HDS 463 HUOJ25 KWBC 091200 /mAVN
> > > 1114c1115
> > > < HDS 201 HUJK20 KWBC 091200 /m93
> > > ---
> > > > HDS 201 HUJK20 KWBC 091200 /mAVN
> > > 1485a1487
> > > > HDS 630 HOMK85 KWBC 091200 /mAVN
> > > 2287a2290
> > > > HDS 909 YRQJ70 KWBC 091200 /mAVN
> > > 3083a3087
> > > > HDS 317 SDUS97 KTSA 091712
> > > 3322a3327,3334
> > > > HDS 310 SDUS99 KWBC 091720
> > > > HDS 311 SDXX99 KWBC 091721
> > > > HDS 312 SDUS99 KWBC 091723
> > > > HDS 313 SDXX99 KWBC 091721
> > > > HDS 314 SDXX99 KWBC 091722
> > > > HDS 315 SDUS99 KWBC 091723
> > > > HDS 316 SDXX99 KWBC 091722
> > > > HDS 317 SDXX99 KWBC 091723
> > > 3387,3392d3398
> > > < HDS 503 SDUS42 KTAE 091723
> > > < HDS 505 SDUS46 KSGX 091724
> > > < HDS 506 SDXX99 KLKN 091722
> > > < HDS 507 SDXX99 KWBC 091721
> > > < HDS 508 SDXX99 KBYZ 091724
> > > < HDS 509 SDUS99 KWBC 091723
> > > 3445d3450
> > > < HDS 943 SDUS97 KBIS 091730
> > > 3480a3486
> > > > HDS 276 SDUS97 KSGF 091730
> > > 3503,3507d3508
> > > < HDS 395 SDUS97 KDDC 091733
> > > < HDS 396 SDUS97 KLWX 091726
> > > < HDS 397 SDUS51 KLWX 091726
> > > < HDS 398 SDUS97 KTSA 091735
> > > < HDS 399 SDUS97 KBIS 091735
> > > 3679a3681,3683
> > > > HDS 283 SDUS97 KLBF 091733
> > > > HDS 284 SDUS97 KAMA 091725
> > > > HDS 285 SDUS97 KEAX 091740
> > > 3865a3870,3876
> > > > HDS 838 SDUS42 KTAE 091753
> > > > HDS 839 SDUS46 KSGX 091754
> > > > HDS 840 SDXX99 KWBC 091752
> > > > HDS 841 SDXX99 KWBC 091751
> > > > HDS 844 SDXX99 KLKN 091752
> > > > HDS 846 SDXX99 KBYZ 091754
> > > > HDS 847 SDUS99 KWBC 091753
> > > 3958,3959d3968
> > > < HDS 645 AGUS43 KUNR 091802
> > > < HDS 650 AGUS46 KPQR 091801
> > >
> > > At this point, it does not appear that the Lone SOH messages in the
> > > log files is significantly different between the two versions of inge.
> > > Thelma is running up to date at present, but it will be interesting to
> > > see if the latencies reoccur today.
> > >
> > > Chiz
> >
> > --
> > Chad W. Johnson E-mail: address@hidden
> > Programmer/Meteorologist Voice: (608) 265-5292
> > Space Science and Engineering Center Fax: (608) 263-6738
> > University of Wisconsin -- Madison
> >
>
> ===============================================================================
> Robb Kambic Unidata Program Center
> Software Engineer III Univ. Corp for Atmospheric Research
> address@hidden WWW: http://www.unidata.ucar.edu/
> ===============================================================================
--
Chad W. Johnson E-mail: address@hidden
Programmer/Meteorologist Voice: (608) 265-5292
Space Science and Engineering Center Fax: (608) 263-6738
University of Wisconsin -- Madison