This is a bit hairy to explain, so bear with me...
We have systems where multiple interfaces are defined on a subnet. This
seemed to work o.k. for Multinet; however, we have now migrated to UCX
and UCX introduces some serious reliability problems.
If one interface on that subnet should lose IP connectivity for any
reason (unco-ordinated network changes, cable damage, etc.), access
to/from the entire subnet becomes so unreliable as to be entirely
useless. Production goes down.
Are we doing something wrong?
Can UCX's "round robin" technique be defeated? (Pointing the default
route to the "primary" interface for the subnet doesn't seem to work as
expected.)
The version, BTW, is V5.4 ECO 4. (No, I've not heard of this being fixed
in ECO 5, at least not yet.) It's running on VMS V7.3-2 using Gig-E on
Fibre.
--
David J Dachtera
dba DJE Systems
http://www.djesys.com/
Unofficial OpenVMS Hobbyist Support Page:
http://www.djesys.com/vms/support/
Unofficial Affordable OpenVMS Home Page:
http://www.djesys.com/vms/soho/
Unofficial OpenVMS-IA32 Home Page:
http://www.djesys.com/vms/ia32/
Coming soon:
Unofficial OpenVMS Marketing Home Page
WOW!!! Almost two months and nary a response!
Maybe I should start pressing my site to revert back to Multinet until
UCX gets this sorted out.
--
David J Dachtera
dba DJE Systems
http://www.djesys.com/
Unofficial OpenVMS Marketing Home Page
http://www.djesys.com/vms/market/