LCAP/802.1ad Trunk

[SKYNHEDD]

Hi folks.

Ive got an LACP link/Trunk that will only ever use three out of the four links.

Has anyone ever had this before?

I can pull cables and stuff, and momentarily it load balances accross all four after plugging back in, but goes back to just three a few seconds after.

Also the un-used link is not always the same one.

Its really annoying as its restricting bandwidth to a SAN.

 

[SKYNHEDD]

Ok

This is what happened. It was working until 2 things happened. I reinstalled the SAN and updated the switch firmware.

Since then no go. Works FINE with a borrowed 3com switch, take the four cables from the existing switch and plug into the borrowed one, works a charm.

Got a replacement 1810-24G from HP, and this fails in the same way. Only three links used.

Sigh... So ive bought brand new HP switch, and if this does the same then Im getting rid of both 1810 switches.

Its ridiculous. The algorithm on the HP cannot be set. All I know is that it used to work perfectly until reinstalling StorageServer 2008R2 on the SAN and doing the switch firmware.

It cant be the SAN, as it works fine when I swap the links to a 3com, with an LACP trunk configured.

Even downgrading the switch firmware back to the original doesnt work, so part of the firmware flash must have changed somethign that cannot be reversed.

Its seriously annoying.

There is updated 1810-24G firmware on the HP FTP site, which is much later than the one for download on the HP driver site, but someone said that this HP support site is password protected.

Does anyone have access to it here, that could get me the P2.5 firmware to have a go with to see if it fixes the issue?

Its literally consuming my life, as I have tried nearly every possible configuration option with the NIC settings and Switch. Since the switch decides where the packets are routed for load balancing, it surely must be the 1810-24G at fault?

 

[SKYNHEDD]

Seriously if someone can help me fix it, I will give them a gift of some sort!

I can draw up a diagram if needed.

Basically I have two ESXI hosts with 2x1gb NICs in Round Robin MPIO connected to the switch just for iSCSI.

The SAN then has the HP teaming sofware on it, aggregating 4 x 1GB links into one 4gbs RX and TX link.

Before I did the firmware and SAN re-install, the packets from each ESXi box were load balanced accross two links on the team. So when uploading to the SAN from the ESxi boxes, all four links were recieving traffic.

Now what happens, is that only 3 of the links ever get traffic.

Sometimes the two Esxi boxes will upload to the same link, so one out of the three is maxxed with packets from both, and the other two are utilised less.

Sometimes the switch will only ever send traffic to one NIC in the SAN team from an Esxi box.. its really really wierd.

All this goes away when using a 3 com switch. I can happily see all four SAN links recieving traffic, and I can pull out cables from the link, and watch everythign auto balance back accross all four links when reconnected.

I just dont get whats happening with the HP !