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

[LDM #QUM-298362]: primary/secondary

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: [LDM #QUM-298362]: primary/secondary
  • Date: Tue, 01 May 2012 13:38:47 -0600
Hi Karen,

> I'm still using the "old" LDM 6.8, so you may not even want to help me.

Wouldn't dream of not helping you. :-)

> I'm running servers within our network as a method of sending our radar
> products to other machines.
> 
> I'm using a "one-stop shopping" idea, in which the majority of our data
> goes to 2 servers and each downstream machine requests from both of them.
> Obviously this is for redundancy.
> 
> However, I'm noticing some problems, in that some of my downstream systems
> don't appear to be getting all of the data due to the flapping.  I can only
> assume that is what it is, since when I request from only one of the
> upstreams I get all of the data, but when I request from both, some data is
> missed.  And sine the 2 servers are so closely located there is a lot of
> switchover going on.  The 2 servers are also fairly busy, so that may be
> part of the issue.

I'm surprised. The LDM system is designed so that switching shouldn't have any 
effect on the products that are received.

Are there any error messages in the log file?

The only way I could imagine this happening is if the REQUEST-s to the two 
upstream LDM-s don't specify the same set of data-products or if the two 
upstream LDM-s don't receive the same set of data-products.

> Is there any way I can actually force a primary and secondary?  So that the
> systems don't switch over unless their primary is actually unreachable?

One can trick the LDM system into believing that two REQUEST-s for the same 
data (to two different upstream LDM-s) actually specify different sets of 
data-products. This will cause each connection to the upstream LDM to be in 
PRIMARY mode -- doubling the bandwidth  and relying on the product-queue to 
reject the second (duplicate) data-product. The trick is to enclose the regular 
expression in one of the REQUEST entries in parentheses. For example

    REQUEST ANY .* ldm1.somedomain
    REQUEST ANY (.*) ldm2.somedomain

The LDM will think that these two requests are for disjoint sets of data 
because the regular expressions don't match. Consequently, it will ensure that 
both connections are in PRIMARY mode.

Does this help?

> Or do you guys have any ideas to insure that I don't lose data to this?
> 
> --
> “Climate is what you expect, weather is what you get."
> -- Robert A. Heinlein
> 
> -------------------------------------------
> address@hidden
> 
> Phone:  405-325-6982
> Cell:   405-834-8559
> INDUS Corporation
> National Severe Storms Laboratory

Regards,
Steve Emmerson

Ticket Details
===================
Ticket ID: QUM-298362
Department: Support LDM
Priority: Normal
Status: Closed