A physical examination of the server should help, but you shouldn't really allow people to drink in your data centre!
But seriously... collect some throughput statistics and see if they are approaching the theoretical limit of the NIC in question. You can do this by running something like ifconfig eth0 or ethtool -S eth0 once every, say, 10 seconds, and calculating the differences between the TX bytes/RX bytes values.
For a 100Mbit NIC, you can devide that value by 10 to get an approximate theoretical limit in bytes per second.
Also check the stats reported by the above for high collision rates and TX/RX errors/retries which may indicate network problems or a misconfiguration (speed mismatch between NIC and port or similar).
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.