NBD setup - No such item

[Tristis Oris]

after telnet request to 10809, rejected packages increased.

Chain xapi_nbd_input_chain (1 references)
 pkts bytes target     prot opt in     out     source               destination
    0     0 ACCEPT     all  --  xenbr0 *       0.0.0.0/0            0.0.0.0/0            ctstate NEW,ESTABLISHED
   17  1020 REJECT     all  --  *      *       0.0.0.0/0            0.0.0.0/0            reject-with icmp-port-unreachable

[Tristis Oris]

if i have no errors during backup, it means XO didn't try to use NBD?

but at begining of each NBD backup i got +1 to iptables rejected list.

[florent]

@Tristis-Oris said in NBD setup - No such item:

but at begining of each NBD backup i got +1 to iptables rejected list.

I think there is something with the network configuration, but I am far out of my element here .

[Tristis Oris]

@florent is it any requirements for network\switch etc. Any layers outside of Xen.

[florent]

@Tristis-Oris it hae to let the encrypted traffic flow through the 10809 port. XO will connect directly to the hosts through NBD, not only to the master

[Tristis Oris]

@florent sounds pretty simple. Nothing at my network should block that traffic.

Anyway, if i got rejected packets at iptables - backup task try to connect NBD server. Then it can't do something, so i should get an error at XO, but it not happens.

[Tristis Oris]

No such item task happens when host interface at status none without IP.

[Tristis Oris]

@olivierlambert i hope your vacation WAS good, be we are stuck a bit here

[Tristis Oris]

bump

[olivierlambert]

Re Ping @florent or @julien-f

[Tristis Oris]

once again i configured it on fresh, test cluster, and it working.

so it some problem with iptables on production. But it default, same 8.2.1 clean installation.

Both this hosts have only 1 connected link, so it both backup and management.
On prod i try empty default backup network, or NBD link network - no effect. So maybe it still going through managemnt link without NBD?

aaaaand enabled NBD on Mng link - now it working. SO yes, it ignore specified backup link.
Or this option didn't work, because NBD now enabled only on Mng link, so it can't go through backup link.

[Tristis Oris]

next weird thing.
i enabled NBD on all manage interfaces at all pools.

it working with CR backup, but shouldn't, at least because it no option to enable NBD.

it working with half of other delta backup tasks, but not for all. Main backup is still without nbd.

only idea, it probably because of 1 small pool with few vm, where i forget to enable NBD. so 1 VM without nbd, force it to work in usual mode?

[florent]

@Tristis-Oris said in NBD setup - No such item:

next weird thing.
i enabled NBD on all manage interfaces at all pools.

it working with CR backup, but shouldn't, at least because it no option to enable NBD.

it working with half of other delta backup tasks, but not for all. Main backup is still without nbd.

only idea, it probably because of 1 small pool with few vm, where i forget to enable NBD. so 1 VM without nbd, force it to work in usual mode?

that is really puzzling. I don't forget this issue, but I don't have much idea of how to fix this

Did you add the preferNbd flag to the config files ?

[Tristis Oris]

@florent lets a brief summary.

nbd work only via management interface. If nbd enabled on any another interface and specified as pool backup interface, it don't work (the issue with blocked packets at iptables). That was a main reason of my problem.

For now i made it work at all my pools. Max single thread speed increased 80-90 > 140-150. At usual backup task (when daily delta very small) it show same 10-40Mb, because it done too fast. Average task time decreased fo about twice.

[florent]

@Tristis-Oris said in NBD setup - No such item:

@florent lets a brief summary.

nbd work only via management interface. If nbd enabled on any another interface and specified as pool backup interface, it don't work (the issue with blocked packets at iptables). That was a main reason of my problem.

For now i made it work at all my pools. Max single thread speed increased 80-90 > 140-150. At usual backup task (when daily delta very small) it show same 10-40Mb, because it done too fast. Average task time decreased fo about twice.

That is really strange, but it is also useful. We'll try to see why , and update the doc accordingly

[Tristis Oris]

@florent also it works too good.
i don' used advanced mode for CR backup, so NBD is disabled for job.

but task is working with nbd)

[Tristis Oris]

that log appear again, without any actions from my side.