@semarie This pool is still on 8.2.1, we are trying to add this host in order upgrade with little to no downtime.

Update 4:

Hello,
After a few tries, i reattempted to re-add the host to the pool, this time capturing the logs in another terminal and this time it seems to have worked.
I think i followed the same steps the previous time, but maybe i did something different, like using pif-scan to introduce the reordered interfaces.

This is how I've managed to get it to work:

All hosts on version 8.2.1 all up to date.

On the new host disconnected the management, remove pif, reorder the interfaces, reboot, add interfaces manually (used pif-introduce, not pif-scan), reboot, re-enable management, join pool.

ifconfig eth2 down
ifconfig eth3 down
xe pif-list device=eth2
xe pif-list device=eth3
xe pif-forget uuid=9104e54c-6c82-5d83-b9fc-1d2b73d5d6f1
xe pif-forget uuid=27abec81-6861-796e-9abe-3b2653444c8f

interface-rename --update eth6=14:23:f2:24:5a:80 eth7=14:23:f2:24:5a:81

reboot now

interface-rename --list
ifconfig eth6 up
ifconfig eth7 up

xe host-list

xe pif-introduce device=eth6 host-uuid=edfaf68e-2c28-4486-8939-723bf2c72820 mac=14:23:f2:24:5a:80
xe pif-introduce device=eth7 host-uuid=edfaf68e-2c28-4486-8939-723bf2c72820 mac=14:23:f2:24:5a:81

reboot now

I'm still baffled why it worked now, I've tried a few time with all the host up to date, and was planning to source a dual nic pcie card, to solve this issue. However on this host the the bond id doesnt match the pool ...
on poll:

on the new host:
xensource_error_log.txt

xensource.log also presents lots of errors, so I'm going to leave it running over the weekend, and will try to add the necessary data for our SR on monday.

what is the best way to send the full log file?

The issue seems to be synchronizing the network configurations.

The pool management is configured for bond1 (eth6+ eth7), that is set for LACP, both switch ports are already configured for LACP.

Bond0 (eth0 +1) is created, but bond 1 fails, and management interface is never assigned to Bond1.

Network 2 and 3 are active on the pool even if never used.

Update 3:
On a fresh install, after renaming eth2 and eth3 to eth6 and eth7 in order to match the active interfaces on the pool, and joining the pool, something happens and the new host dom0 loses the networking interfaces.
Trying to do an emergency network reboot does nothing, the network doesn't come up, and doing any xe command returns nothing.

Seems like not having 2 interfaces to match all 8 interfaces on the other hosts is an real issue.

Any feedback or recommendations are welcome. The docs don't cover this, but is it possible to delete unused networks associated with non-used PIF?

Update 2:
Renaming eth2 to eth6 and eth7 worked like a charm.

Had to do a fresh install because the emergency network restart didn't work.

it's hard to tell which PIF belong to each bond on XO (at least on source-compiled version), needed to use old xcp-center to figure it out.

This thread can be closed.

still having some network issues :S

Update:
I'm having issues resetting the network on the new host, but looks like the issue is eth6 and eth7 are the fiber ports.

I'm going to reinstall the new host and reorder the eth2 and eth3 to eth6 and eth7, then try to rejoin.

@DustinB The issue is lack of pool resources to have all prod machines up, if i can't join the new host withou update, i'll have to schedule some downtime.

Hello there,

We have a production pool (2 hosts) that is in dire need of a new host, so we removed an newer host from another pool, but we stepped on something that i knew could be an issue.

The production pool has 2 hosts on xpc-ng.8.2.1, that had issues updating to 8.3, each host has 8 PIF interfaces, 4 copper 1gbps, 4 fiber 10Gbps. 2 of the copper PIF are not used. The newer host only has 6 PIF, 2 copper 1gpps and 4 fiber 25 Gbps.

I made sure the PIF order on the new host matches the order of active PIF.

When i try to add the new host to the pool, the dom 0 on the new host stops having network access, and shows no PIF. Configure Management Interface reports <no interfaces present>.

I bet the issues is the PIF mismatch, is there anyway i can disable the unused PIF and get add the new host? Can it be a software version missmatch? we didn't update the new host after install, because the pool is one patch behind.

Bonus question, is there a way to install guest tools on older Operating systems, like windows XP and centos 4, for some old guest systems we need to keep arround ... (yeah i know we should get alternatives ...)