HP Procurve 1810-24G incompatible with VE 3.x

[jcheger]

I did migrate my 2 servers, and had quite a bad surprise. Corosync was failing after less than 5 minutes. After many hours of investigation, I did discover that the Procurve 1800-24G switch was the cause of the problem. A multicast test using asmping did show 25% packet loss in multicast. Those 2 servers have never been unsynced with Debian Squeeze / Proxmox 2.x. Moved on a 1810-24G, and it the problem is magically gone.

 

[tom]

if you use multicast on managed switches, you need to configure them. out of the box, multicast traffic is blocked by a lot of enterprise switches.

 

[jcheger]

if you use multicast on managed switches, you need to configure them. out of the box, multicast traffic is blocked by a lot of enterprise switches.

On a lot, but not on Procurve 1800 / 1810. Multicast is allowed out of the box, without any option. Although we can configure some limitation (1800: rate limit / 1810: storm control), that I don't recommand anyway, because many other broadcast paquets are also dropped, i.e. Bonjour.

I couldn't find any message about this issue on internet, related to the Procurve 1800. Now there is one.