T.R | Title | User | Personal Name | Date | Lines |
---|
9000.1 | | SMURF::MENNER | it's just a box of Pax.. | Fri Feb 28 1997 19:33 | 4 |
| Some software uses local loopback as a form of inter process
communication. Since the loopback is not a real network interface
you can increase the MTU. Larger MTUs decrease the change of
fragmenting TCP/UDP data.
|
9000.2 | changing the defaults? | VIRGIN::SUTTER | Who are you ??? - I'm BATMAN !!! | Mon Mar 03 1997 03:54 | 7 |
| If this is generally a good idea (and many system check utilities like
UNIcensus suggest that this parameter is beeing raised) why don't we
change the default to a reasonable value?
Regards,
Arnold
|
9000.3 | | SMURF::MENNER | it's just a box of Pax.. | Mon Mar 03 1997 07:56 | 3 |
| Re: .2
good idea!
|
9000.4 | I give you the permission to do it ...!... ;-) | VIRGIN::SUTTER | Who are you ??? - I'm BATMAN !!! | Mon Mar 03 1997 09:54 | 7 |
| Ok, let's do it then.
Who to contact/involve in such a major decision?-)
Regards,
Arnold
|
9000.5 | | DECWET::ONO | Software doesn't break-it comes broken | Wed Mar 26 1997 14:57 | 12 |
| The mtu recommendation of 16384 was posted in the NetWorker
conference a while back. A customer recently upgraded to DU
4.0b, and this value caused his local IP throughput to drop
dramatically. Setting mtu to 15000 improved things, and he says
his current value of 3072 gives the best performance.
I'm trying to figure out why the customer sees this behavior.
Regards,
Wes Ono
NetWorker Engineering
|
9000.6 | | DECWET::ONO | Software doesn't break-it comes broken | Wed Mar 26 1997 14:59 | 3 |
| additional info re: .5
The 16384 value worked fine for 3.2g.
|
9000.7 | get a tcpdump trace | SMURF::DUSTIN | | Thu Mar 27 1997 08:59 | 7 |
| The best way to find out what's going on is to get a tcpdump
trace of the traffic on lo0. Starting in the V4.0 release,
you can now run tcpdump on the loopback device, which makes
the problem much easier to diagnose.
John
|
9000.8 | | DECWET::ONO | Software doesn't break-it comes broken | Thu Mar 27 1997 20:29 | 7 |
| John,
We're working on this (as I mentioned in response to your mail).
Thanks for the assist.
Wes
|