Environment:
Solaris Master Server (Netbackup 4.5 MP5) multihomed with a primary 10/100
and a Gigabit NIC on a dedicated subnet.  All clients are windows based (NT
and the all varying flavors of 2000 except datacenter)  All clients have
their own policy.

Over a 24 hour period and out of 75 machines, roughly 7 will generate a 174
error and fail the job mid-backup.  In the next 24 hours, another set of
machines will fail, but it varies in number of machines...very intermittent
and random because the same machines aren't always the same. 

There are a couple simularities ....
1)  The error  "Attempted to write ### gigabytes, ########## bytes, it is
not a multiple of 512." appears on all the machines that generate a 174.
(the ## are all different values from machine to machine.)

2)  All the machines are on the backup only subnet.  No machines that use
the production LAN as their path generate this error.

I found this technote: http://seer.support.veritas.com/docs/262860.htm , but
as the gigabit interface on our box is in a Sbus (25mhz) slot and we have no
PCI's available, I'm sure that this related to our issue but don't know how
to resolve it.

Anyone have any thoughts?

Christopher Johnston
Windows NT/2000 Platform Support Group, MCP CCNA 
Aegis Communications Group, Inc. - Irving, TX