Quick comment on this before I crash - there was discussion
on an opensolaris.org forum (network-discuss? nov-dec'06?)
about similar (same?) issues, and adding the ability for a process
to require a specific ipaddr (or interface?) be used for outgoing
traffic (like a per connection sockopt for example).
As opposed to choosing any available route
back to the original source addr.
Don't recall the details off-hand nor the decision if any,
but if it's the same issue then it ain't in yet (I recall the
impression it would be regarded as a useful feature [RFE]
rather than a bugfix as such).
In (almost) all cases we now use IPMP with interfaces
marked as standby so only one is active at a time,
although it's not ideal. Trunking (aka Etherchannel)
might be another way around it - this is at MAC layer,
all interfaces in group have same MAC/ipaddr's,
but it has other restrictions (eg. can't group ether+atm,
which you can do with IPMP if you really want to).
Rgds, Stuart.
Stuart Remphrey
RMIT ITS Infrastructure Services - Unix Systems
Phone (03) 992 55 070 (or extension 55070)
>>> <chrisj@ucia.gov> 02/03/07 10:53 PM >>>
...
Solaris does indeed use the multiple interfaces
in a seemingly random way (well, it really isn't
random ... I'm sure there is a very logical approach
but to the occasional telnet, ssh, scp, or NFS
application it appears random :-).
Carson pointed out that it might be a solaris bug
for which a patch already exists.
...
|