Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

NIC Collisions

Status
Not open for further replies.

allywilson

Technical User
Nov 27, 2002
157
GB
Hi all,
We have 3 Solaris boxes (5.10) that are reporting massive amounts of collisions on the NICs.

Server1 info
NIC (/dev/eri) set to 100 autonegotiate - currently half-duplex.

Server2 info (/dev/bge0) set to 100 autonegotiate - currently full-duplex.

Server3 - same as Server2.

When I try to force Server1 to use full duplex
Code:
# ndd -set /dev/eri adv_100hdx_cap 0
# ndd -set /dev/eri adv_100fdx_cap 1

It results in
Code:
# ndd -get /dev/eri link_mode
# 0

I don't want to block half-duplex and stop it negotiating (as I think it will stop all communication with the switch).

But the half-duplex issue doesn't explain the other 2 servers.

Any ideas?

Oh, and the reason I'm investigating all of this is that it's taking >2mins to transfer an 8Mb file between Servers 2 and 3 with 1.
 
Is there any way to check the switch settings from Solaris ?

I have a similar problem, the server is hardcoded to 100/full on the backup port ( for backups ) we noticed that backups were slow and it finally turned out the switch was at 100/half.

Is there anyway we can check this from Solaris ?
 
I don't think so because if there were, it would spit up a mismatch error message!
 
Guys, thanks for the input.

Got a hold of one of the network/cisco guys. It seems he was aware of a flaw with the particular IOS we were running on the switches which if it was hardcoded 100 full duplex on the switch side and set to auto on the server side sometimes resulted in a false positive (both switch and server reported full [exception was Server1 - which correctly detected half] but were in-fact half). He modified the config to be auto negotiate, et voila. Am back up and running at 100 full on all 3 switch ports.

Good times :)

Is there anyway we can check this from Solaris ?
Without sounding like I'm being condescending, I guess the only way would be to telnet to the switch from the server.
 
How do we check if the server is communicating at 100/full ?

In my case dladm show-dev, ndd all showed that the port was at 100/full. netstat did not show any errors the only indication was slow backups.

Is there anyway to trouble shoot such issues without involving the network guys ?
 
No. You can only tell what you negotiated, you cannot tell what the settings on the switch were without knowing how the port was setup on the switch side.

Now if the network guys say "all the ports are X speed/duplex or auto" then you probably can make some deductions on mismatch, but without that initial info, you can't.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top