ARP Table refresh in 4000m and 2524

[Marc Villeneuve_1]

Hi.
I would like to know what is the refresh time on procurve 4000m and procuve 2524 for the ARP table. We have a weird probleme with our cluster server and we suspect the switch... but not sure.

TX
Marc

[Jim Smith_19 · Accepted solution]

Could you provide more details? These are layer 2 switches and should only use their ARP tables for personal use (switch management, etc.) If you are looking for an ARP cache problem between two other nodes whose traffic is passing through the switch, this should not relate to the switch's ARP table but to a problem with one of the nodes. Hope this helps.

Jim

[Marc Villeneuve_1]

Ah great. Tx for the answer. This is what I was looking for! I was sure of this answer, but I was seeking someone to confim this. It must be the server, after a patch, who doing the problem. I know that the switch at this layer(not a routing switch) not doing anyting with arp with other devices, but my consultant was not sure.

Since I have post this question. My net admin look farther more and found a patch for tcpip. We think we resovle the problem but not sure.

Tx again.

Marc

[Les Ligetfalvy]

While Jim may have been right in this case, I still take exception to his statement. Layer 2 switches DO rely on ARP (MAC) tables to "route" through the fabric. That is the only way they know what path to take. Every port in the fabric holds the MAC of devices on that port and the switches periodically refresh these tables. IIRC the schedule is somewhere around 15 minutes.

I once had a badly behaving switch that was misrepresenting a MAC address and it kept subverting the route through the fabric. The short term workaround was to continually ping the address so as to lessen the effect of the bogus table until I could schedule the downtime to change out the switch.