I wish that one day, just one day, this would work right. Specified backup network NIC
-
@elementalwindx what network can the xoa access? can you show a screenshot of the network tab of xo ?
are you using a proxy ?are you doing backups or replication ?
-
-
@elementalwindx I think I know : the backup network is used for the host storage to xo path , not for the xo to replica (also host)
that is why you only have RX , and not RX + TXedit ; after checking the code, it should also be used when writing to an host
-
@elementalwindx yes XOA need to have access to the backup network . the default migration network is used directly by xcp-ng no need to assign to xo
-
we ran into this issue too.
By default (coming from vmware) we used management network on 1Gb NICs.
we have 2x10Gb and 2x25Gb nics also in the hosts, and struggled to get backup on THESE nics...
we had to :
- assign a nic and IP to XOA int our preferred backup network (we created a dedicated VLAN on 2x10Gb bond)
- be sure that each host has an IP address in this network too
- we added XO Proxies... guess what ? they also need an IP address in this network (on the FIRST nic if you have more than one per proxy... the first IP address reported by XAPI on the proxy or xoa default tab must be the backup network one... haha.)
and then, put the correct preferred backup network on the pool, and it takes the good nics to transfer backups. hooray.
ps : beware, when all is working as intented... check your STATS tab... we had half our pools/hosts not reporting anything else than NO STATS, and the regular RRD scan task was failing after changing this BACKUP NETWORK preference !
Why only half ? all host/pools that were in ANOTHER subnet of the XOA vm weren't reporting stats anymore.
we had to add static routing to get the stats tabs work again, the XOA needed to still contact hosts that were on another subnet.whats the correlation between RRD scans to make stats and prefered backup network ? don't ask me
.
-
@Pilow you found one of the not so bright part of the api
the data path is the same for backups and RRD, so RRD failing are often a canary of a misconfigured backup network. So if you change backup network and the task are falling, this probably means something wen wront with your network.
-
@Pilow HOLY COW. Could they make it any more unnecessarily more complicated? shesh.
-
@elementalwindx it seems complicated, coming from other virtualization stacks that has billions of R&D budget with solutions where you click next next next, done.
but in the end, when you understand the mechanisms, and get it working, you really fully understand whats going on.
many facepalms in the process though
-
@Pilow yeah I think after 4 years of dealing with it, I finally give up. Been moving everything to proxmox for the past couple days.
-
@elementalwindx what works for you is the best !
we are setting up a full.Vates.stack MSP hosting solution.
xcpng has its current flaws, but they are addressed each month by the dev team -
different platform have different internal , and so different edge case. We are here to help to bridge the gap.