LAG Network connectivity fails when adding iSCSI interfaces
When I assign a new iSCSI network interface, the management network interface stops responding to pings and won't come back up until I reboot. Also, when initially adding the LAG I have sometimes needed a reboot in order to get it to come up properly and accept a DHCP lease.
Steps to reproduce:
- Install up new FreeNAS install (latest version FreeNAS-9.3-STABLE-201509282017) with 4 supported network cards
- Configure Link Aggregate Group for 2 network cards
- LAG does not come up, reboot, LAG comes up, DHCP lease accepted
- Log into WebUI
- Add new network interface on separate iSCSI subnet with MTU 9000
- Dialog stuck on 'Please wait...', WebUI stops responding, LAG stops responding to pings, Switch reports LAG channel Active on both associated ports
- Reboot, LAG comes up, iSCSI interface works
- Add second iSCSI interface, reboot, etc.
I have reproduced this issue about 5 or 6 times on 3 different sets of hardware. I can provide whatever logs you'd like.
#8 Updated by Thomas Maerz almost 5 years ago
The system I am accessing the GUI from is on the same subnet as the LAG (management interface). The iSCSI subnets are not accessible to the rest of the network or each other. So I have
LAG: 10.0.50.0/24 <--This is the only place I access the GUI, and that subnet is accessible from my management machine.
iSCSI 1: 10.3.0.0/24
iSCSI 2: 10.4.0.0/24
#9 Updated by Josh Paetzel almost 5 years ago
Ok, well that blows away my theory of a route is getting lost.
So, riddle me this.
Do you have remote console on these machines?
If yes, could we do a teamviewer session where you can get the FreeNAS system into the broken state where I can troubleshoot what is wrong on the remote console?
#10 Updated by Jason Brooks almost 5 years ago
I can confirm this bug. Please see https://forums.freenas.org/index.php?threads/link-aggregate-group-stops-working-when-adding-other-new-network-interfaces.39574.
rebooting the system builds the lagg interface correctly.
seen in FreeNAS-9.3-STABLE-201511280648
#11 Updated by Josh Paetzel almost 5 years ago
I need access to a system that is exhibiting this behavior. Preferably one with IPMI or serial access so I can see it when it's in it's broken state.
One of the challenges here is I use this config at home (LACP + MPIO for iSCSI) and I am unable to reproduce the problem.
#13 Updated by Jason Brooks almost 5 years ago
I can do this: it does it on a test system as well as my live system. I will need to address this monday though: I am about leave for the weekend.
I have had some trouble with directly connecting freenas to the ipmi device, but I think I can at least network you into it...After all, I am also the network admin... :)
#15 Updated by Thomas Maerz over 4 years ago
I had a mirrored USB failure and experienced this again while setting the system back up. This time I tried configuring the LAGG from the GUI. Same result. I can't show the system in a broken state because this system only has iDRAC express which doesn't have an IPMI remote console feature unless you get iDRAC Enterprise licenses.