Long Cable length causes failure on port self test?

[Kevin Wigle]

We are moving into a new building. We don't have control over the patching/infrastructure. We have a 3500 switch in the data centre and from the switch I can ping all over. My new office (bullpen) is on the second floor with 3 other co-workers. In the wire closet we have a 2824 switch. The two are connected via copper and a wire measurement says something on the order of 154 meters and I'm told there are at least 4 patches to get from data centre to our bull pen.
The thing is - I cannot ping from the data center to the wire closet. (154 meters) Obviously this is more than spec 100 meters. However, if I replace the 3500 with a Cisco switch - it works. All switches have bare configs, just the default vlan and IP address.
When using the 3500, the link light flashes about once every second (or slightly more) I have googled around and it appears that this is an indication of the port failing its POST.
I can move to any port and the result follows.
While the one port is failed all others ports are working just fine.
However, plug the Cisco into the very same port and it works.
So..... is the Cisco able to manage longer cable runs? (if unofficially)
Is the HP sensitive to long cable runs, does a test and fails the port?
Or another idea?

[Richard Brodie_1]

It might be. Another possibility is that it is auto negotiating a different speed. If you have Gigabit ports at both ends, less likely to work than if one is at 100Mbit/s.

If you have ~150m of cat 5, it is likely to work fairly reliably at 10Mbit/s. I would try setting the 2824 to auto-10 (or maybe auto-100). Worth keeping an eye out for packet loss, however you leave it.