Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Wed, 26 Sep 2007 05:29:24 -0700, Kelly Byrd <kbyrd-drbd at memcpy.com>
wrote:
>
> I'm no CentOS 5, using the drbd packages from them (8.0.4 I believe)
>
> The beginning of my drbd.conf contains.
>
> |common {
> | syncer { rate 200M; }
> | protocol C;
> |
> | startup {
> | degr-wfc-timeout 5; # 5 seconds.
> | wfc-timeout 5;
> | }
>
> On a machine boot, it seems the drbd init.d script wants to
> wait forever because of the call to 'drbdadm wait-con-int'
> The output from that syas the degr-wfc-timeout and wfc-timeouts
> are '0'.
>
>
> What's going on?
>
>
I may have answered my own question. Searching the archives,
I found this from Lars:
| why not for a clean reboot of a standalone Primary?
|
| first, because _every_ clean reboot looks the same way:
| become secondary, disconnect, unconfigure, reboot.
| so we have been "disconnected" (degraded?) before the reboot.
|
| second, because if you do a clean reboot,
| aparently you are at the box anyways,
| so you can interactively interrupt the "wait-forever".
Not true. I'm ssh'd into both nodes doing administrative work.
I cleanly bring down the drbd services, do my thing, then reboot
them with 'reboot', intending the boxes to come back into service
or at least let me ssh in after they reboot. I'm not physically
at the box.