[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: NOAAPORT receive system feeding LDM

On Tue, 19 Oct 1999, Dan Vietor wrote:

> > Does the unisys NOAAport systems use WMO headers?
> 
> Yes, it produces a standard WMO header and removes the NOAAPORT header.
> 
> SOH,CR,CR,LF
> seq,CR,CR,LF
> WMO header....
> 
> and adds a terminating ETX,LF on the end of a product if none exists.  I
> don't believe we put CRs after the ETX but I can check.  Those pesky
> binary products.
> 
> > Also, Does the products have a sequence number.
> 
> Yes, but we start the sequence number at 001 when the ingest process
> starts and increment to 999 and recycle.
> 
> BTW, do you have the algorithm that SSEC uses for the sequence number.
> We can add it here so we can remove duplicates between our feed and IDD.
> 
> > The bottom line, does the unisys NOAAport stream look like the
> > SSEC NOAAport stream?
> 
> Having not looked at the SSEC feed, I can't say for sure.  I was able to
> successfully ingest the feed on LDM 5.0.8 here for about a 72 hr period
> but I'm not sure that is long enough for a thorough test.  I'm
> contemplating doing another test here but securing systems to do it here
> is not easy.
> 
> > The question that I'm concerned about is rejecting duplicate products.
> 
> I think this is Jim's major problem.  He is getting product from his
> Alden receiver, IDD and our NOAAPORT so in many cases he is getting
> triplicates since the WMO/seq headers don't match.  This means his
> product queue has to be enormous.
> 
> I am curious what would be giving Jim a "permission denied" to write to
> the product queue.  This terminates the pqing process.
> 

Dan,

I don't understand why you are getting the permission denied error. We
have NOAAport card being developed here at UPC that is receiving all 4
channels without any errors.  The only thing that comes to mind is the
rate that the products are being inserted. But, our system sends the
products straight from the card to the pqing without problems.  The queue
is large $pq_size = 800000000; 

SunOS jackie.unidata.ucar.edu 5.7 Generic_106542-05 i86pc i386 i86pc

I wonder if the unisys system would run on solaris?


Robb...

 Here's the section of LDM code pq.c, producing the error:

        /* get the actual data region */
        status = rl_r_find(pq->rlp, tqep->offset, &rp);
        if(status == 0
                 || rp->offset != tqep->offset
                 || Extent(rp) > pq->ixo - pq->datao
                )
        {
                char ts[20];
                (void) sprint_timestampt(ts, sizeof(ts), &tqep->tv);
                uerror("Que corrupt: tq: %s %s at %ld",
                        ts,
                        status ? "invalid region" : "no data",
                        tqep->offset);
                /*
                 * we can't fix it (tq_delete(pq->tqp, tqep)) here
                 * since we don't have write permission
                 */
                status = ENOERR;
                goto unwind_ctl;
        }             





> > Are there any other major differences that I should be aware of?
> 
> Not that I know of.  If you want, I can forward some raw files that I'm
> feeding to pqing.
> 
> > Thanks,
> 
> Let me know what you need and I'm be glad to help out.  If I get that
> test system set up, I can follow up on the problems here.
> 
> Thanks!
> 
> Dan.
> 

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================