[DRBD-user] Testing DRBD9*.rpms: Not able to join the second node

Roland Kammerer roland.kammerer at linbit.com
Wed Dec 28 10:38:14 CET 2016

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


On Tue, Dec 27, 2016 at 10:17:32AM -0600, T.J. Yang wrote:
> Hi
> 
> I am testing out latest DRBD9 rpms, hoping to create bravo,alpha and
> charlie 3 nodes cluster. Need pointer on where I did wrong.
> 
> 2.  Trying add 2nd node alpha from bravo node.
> 
> drbdmanage add-node alpha  192.168.174.136
> 
> The command failed after waiting on server (which server ?)

Its a client/server application, the server is drbdmanaged.

> I login into alpha to run the command manually.
> 
> [root at alpha drbd9-rpms]# /bin/python /usr/bin/drbdmanage join -p 6999
> 192.168.174.136 1 bravo 192.168.174.141 0 AMcP2OIcqCwO+iTer5Bx

That was spit out by the failed autojoin or by "howto-join", or manually
crafted?

> Waiting for server: ...............
> Error: Server currently not ready, please retry later

drbdmanaged failed to start up, because it failed on DRBD level.

> [ 9031.805838] drbd .drbdctrl alpha: Starting receiver thread (from
> drbd_w_.drbdctr [42225])
> [ 9031.806840] drbd .drbdctrl alpha: conn( Unconnected -> Connecting )
> [ 9031.807425] drbd .drbdctrl tcp:alpha: bind before listen failed, err =
> -99
> [ 9031.808002] drbd .drbdctrl alpha: Failed to initiate connection, err=-99
> [ 9031.808551] drbd .drbdctrl alpha: conn( Connecting -> Disconnecting )
> [ 9031.811144] drbd .drbdctrl alpha: Connection closed
> [ 9031.813713] drbd .drbdctrl alpha: conn( Disconnecting -> StandAlone )
> [ 9031.814300] drbd .drbdctrl alpha: Terminating receiver thread
> [ 9166.872047] drbd .drbdctrl alpha: Terminating sender thread

That is the real issue. DRBD creates a kernel socket and calls
sock->ops->bind(). This returns -99, which is -EADDRNOTAVAIL. So the
question is why wouldn't it be available? Everything right with your IP
addresses? If you have multiple you used the one you want to use for
cluster communication on "init"? The nodes can reach each other and
everything is nice and cozy from a network point of view?

> [root at bravo log]# drbdmanage n
> +---------------------------------------------------------------------------------------------------------+
> | Name  | Pool Size | Pool Free |      |
>                          State |
> |---------------------------------------------------------------------------------------------------------|
> | alpha |   unknown |   unknown |      | offline/quorum vote ignored,
> pending actions: adjust connections |
> | bravo |      4088 |      4080 |      |
>                             ok |
> +---------------------------------------------------------------------------------------------------------+

"normal", as the control volumes did not connect.

Regards, rck



More information about the drbd-user mailing list