remsh , it is very sloww

[Clodoaldo João Rodrigue]

Hello All,
We have a problema in our networks, we have four servers applications (L2000 with HP-UX 11i) and one N4000 and one Super Dome witn two partitions, all the machines it is by GB Networks. So when we try to execute a remote command (example: time remsh host date) from Supoer Dome to N4000, it display 0.2, from partition 01 to partition 02, it display 0.2, but when we try to all L2000, it is display 0.5, 0.6 seconds. What´s wrong ?? If all the server are in the same network.
Thanks for all

[Pete Randall]

Usually something like this is a result of a duplex mismatch. Since you're all GB, this seems unlikely, but check each of your servers with the "lanadmin -x 0" (where 0 is the the PPA or instance number of the lan card, found with lanscan). For GB networks, the cards and the switches they are connected to should all be set to auto-negotiate.


Pete

[T G Manikandan]

Try using ip-address instead of hostname in the /.rhosts file.


Revert

[Mark Grant]

Two things I can think of as possbilities.

One is odd routing. Check the output of "netstat -r" and see if some have convoluted routing.

Another thing is to check to see if it is a problem with the interface cards negotiating different values with the switch. We have seen similar problems when some machines are halfduplex instead of full.

[harry d brown jr]

It depends on a lot of things like system loads, security policies, bind and/or routing.

Here is an example of an N4000 partitioned with 5 partitions. All on the same switch - note ==> not a router.

# time remsh vpart3 time

real 0.3
user 0.0
sys 0.0
# time remsh vpart4 time

real 0.2
user 0.0
sys 0.0
# time remsh vpart5 time

real 0.3
user 0.0
sys 0.0
# time remsh vpart6 time

real 0.2
user 0.0
sys 0.0
# time remsh vpart1 time

real 0.4
user 0.0
sys 0.0
#

and vpart1 is the machine I was remsh'ing from!

live free or die
harry

[Michael Steele_2]

I agree. Disable NIC auto and replace with 100HD or 100FD on all NICS including switches. Use 'linkloop MAC' to test for connectivity instead of 'ping IP'. Use 'lanscan' and 'arp -a' for MAC's on the server side, see your network admin for the switch MAC's and speed and duplex. To change speed and duplex locally:

lanadmin -x NMID
lanadmin -X 100FD NMID, for example

lanadmin -x 4
lanadmin -X 100FD 4

[Steven E. Protter]

Once the ladadmin duplex issue is resolved, you may want to make srue the .rhosts files have hostnames that are supported by your DNS server infrastructure.

For some servers you may need to hardcode those duplex settings into the lan configuration file like /etc/rc.config.d/hpbtlanconf.

Here is an example(attached)

SEP