[DRBD-user] Alternative to 'yes' to abort waiting?

[Whit Blauvelt]

On Sat, May 28, 2011 at 02:58:28PM -0700, Matt Graham wrote:

> If you can ssh to the bad machine, fix the /etc/init.d/drbd script so that it
> starts *after* all the NICs are running.  

Yeah, I could do that if I could ssh. If I could ssh all would be pretty.
But the drbd init.d script is blocking networking from starting. ; <

Now, how did it end up that way? Set this and its secondary up many months
ago, in something of a hurry. The version of drbd on them was built from
git at the time. I'd assume the init.d script came in with that, so a
LINBIT thing. In this case it's on an Ubuntu server.

> In all of our CentOS DRBD clusters, the drbd init script is never called by
> init.  init calls heartbeat, and heartbeat does all the DRBD stuff.  Are you
> running DRBD without a cluster manager, or something?

DRBD is running with heartbeat in this case, and you're right that the drbd
init script probably shouldn't be called this way. Although to quote from
the INIT INFO section of the script:

# X-Start-Before: heartbeat corosync
# X-Stop-After:   heartbeat corosync

Does "X-Start-Before" mean start this before these, or start these before
this? Ubuntu as a Debian should obey this LSB stuff. But
http://wiki.debian.org/LSBInitScripts doesn't specify the syntax in much
detail. The drbd init.d does have:

# Required-Start: $local_fs $network $syslog

which pretty clearly should be telling the system not to start drbd before
the network is up. But it's trying anyway. Although that Debian page says:

$network    low level networking (ethernet card; may imply PCMCIA running) 

So "low level networking" might mean something less than actually having
networking running? It could be that this is an instance of tripping over
Ubuntu's "upstart" transitional stuff. Much as I respect Ubuntu Server 
(I've run Slack, Red Hat, Gentoo, Debian, still run CentOS - and prefer
Ubuntu), upstart is a PITA compared to proper init scripts.

Ah well,
Whit