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

[LDM #BIW-247121]: best way to re-insert a product in an LDM queue under a different feedtype



Tom,

Your idea of using pqinsert(1) to change the feedtype is a good one and should 
work.  Use the "-i" option to have the MD5 checksum computed from the 
product-identifier and ensure that all product-identifiers are unique.  This 
will allow the queue to contain the same data-product but with different 
feedtypes.

> We have been having email exchanges with folks at the WOC about the best
> way to include new products into the CONDUIT feed (CONDUIT originates
> on machines at the WOC).
> 
> Here is the body of an email that Patrick O'Reilley sent us yesterday
> about how they are thinking of adding the products:
> 
> Here's a rough dataflow diagram of the path of these data:
> 
> TOC system socket => WOC system socket => WOC socket LDM => WOC CONDUIT LDM 
> => Unidata and others
> 
> where the "socket" is a client/server software.
> 
> When the WOC socket software inserts it into the "socket" LDM, there is a 
> default
> feedtype it is assigned.  There's not any flexibility to assign different feed
> types.  So since in the future, there may be products coming from the TOC to 
> the
> WOC that aren't destined for CONDUIT via this path, I guess it's a good idea 
> to
> assign it to EXP, and make that permanent.
> 
> With this in mind, is there an easy or standard way to change the feedtype 
> between
> LDMs?  After the test, the products will be coming into the WOC CONDUIT LDM 
> as EXP
> and we need them assigned to the CONDUIT feed type.
> 
> We can think of a couple indirect ways but thought there may be a more 
> straightforward
> method.
> 
> Thanks!
> Patrick
> 
> In recap, the issue is that they want to send the new products to the CONDUIT 
> injection
> machines using the LDM EXP feed type, and then have the new products 
> re-inserted into
> the CONDUIT injection machines LDM queues in the CONDUIT feedtype.
> 
> I commented to Patrick that it might be possible to re-insert the same product
> into a queue where the original is still resident by assigning an MD5 
> signature
> to the re-inserted product (assign one, not let pqinsert calculate one).
> 
> Questions:
> 
> - would this circumvent the duplicate product rejection?
> 
> I think that it would, but I wanted to make sure that there were not 'gotchas'
> 
> - if the answer to the first question is yes, it seems to me that the best
> thing to do is to assign the artificial MD5 signature to the product in
> the EXP feed and let pqinsert calculate the signature when the product
> is re-inserted as a CONDUIT product
> 
> Any comments would be most useful...
> 
> Cheers,
> 
> Tom
> --
> ****************************************************************************
> Unidata User Support                                    UCAR Unidata Program
> (303) 497-8642                                                 P.O. Box 3000
> address@hidden                                   Boulder, CO 80307
> ----------------------------------------------------------------------------
> Unidata HomePage                       http://www.unidata.ucar.edu
> ****************************************************************************

Regards,
Steve Emmerson

Ticket Details
===================
Ticket ID: BIW-247121
Department: Support LDM
Priority: High
Status: Closed