Catalyst 2950 -throughput issue

[joyride_us]

Hi,

I have 1 Catalyst 2950. All ports are busy.

on 2 of them I have 2 Unix servers with Ethernet boards configured as 100Mb/FD.

Everybody can ping everybody with a nice throughpout (20-40 Mb/s) except those 2 Unix servers that cannot ftp anything above 4 Mb/s (which is not enough for certains applications the run)...The ports they are connected to, are also configured as 100Mb/FD.

Where could the trouble come from ?

Thank you

Jef

[bazarganc]

have you checked duplex and speed on these ports ?

[joyride_us]

Hi,

thank you for the reply. Yes we have check and tried just about every combination. But since everything is set to 100MB/FD, we assume the problem is not coming from there.

I start to believe that there is an application n the server that is kiiling the queue on the switch port...the hash signs when I fpt a file show very irregular transmit rats like a General synchronization issue due to tail dropping. I will take a look tomorrow!

Thank you

Jef

[Kevin Dorrell]

Can the Unix servers do autonegotiation?

Have you seen this thread?

http://forum.cisco.com/eforum/servlet/NetProf?page=netprof&CommCmd=MB%3Fcmd%3Ddisplay_location%26location%3D.1dd6a4c8

Kevin Dorrell

Luxembourg

[Tsasbrink]

You could check the backplane load with the show controllers utilization command.

Check the specific interfaces for error's and other nasty counters with the show interface command.

If these do not give you answers you should focus on the servers.

hth

Tom

[joyride_us]

Hi,

thank you for the replies but we upgraded the FW of the switch and it is now behaving properly! Spooky stuff!

[Kevin Dorrell]

Glad it's working now. What was the IOS version before and after? Just for reference in case we ever see the symptoms again. I'll also do a bug search on the versions.

Kevin Dorrell

Luxembourg

[joyride_us]

Hi Kevin,

the buggy one : 12.1(13)

the working one being : 12.1(14)

Cheers

Jef