Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi Lars, thanks for your quick response.
> > * startup script for gentoo is somewhat minimalistic- any special
> > reason the script does _not_ load the drbd module?
>
> well. we don't gentoo. I never even looked at that script,
> its been some user contribution.
ACK. I'll post my modifications once I've got it up and running.
> > Kernel is 2.6.19
>
> hopefully not. 2.6.19 is corrupting data...
> 2.6.19.2 might be a better one?
Sorry, omitted the final part of the version. Yes, it's actually
2.6.19.2.
> btw, since you do drbd on md raid, did you notice
> http://thread.gmane.org/gmane.linux.raid/14142/focus=14142
Yes, but I wanted to get started with fewest possible patches and start
adding these once I've got basic functionality up & running.
> it does work with "internal" instead of "flexible".
> to drbdsetup, "flexible" means "flexible external", and that
> correctly won't work if the "external meta data device"
> is the same as the "storage device".
>
> to drbdmeta, unfortunately there is "internal" as well as
> "flex-internal", which mean the same thing for drbd 8, but
> not for 0.7, for compatibility reasons...
>
> > flexible-meta-disk internal;
>
> if it is internal, it does not matter whether you say
> "flexible" or not.
The reason I ended up with this combination is that it's the first
meta-disk example in the sample config file :-)
I'll retry with meta-disk internal ...
OK.. Better: now I get a completely different error:
* Starting DRBD ...
drbdadm: stack smashing attack in function adm_connect()
/etc/init.d/drbd: line 39: 15940 Aborted ${DRBDADM} up
all
I'm using gcc 3.4.6 with built-in stack protection.
Debug output from adm_connect:
argc: 21
drbdsetup
/dev/drbd0
net
192.168.0.23:7788
192.168.0.21:7788
C
--set-defaults
--create-device
--rr-conflict=disconnect
--after-sb-2pri=disconnect
--after-sb-1pri=disconnect
--after-sb-0pri=disconnect
--shared-secret=xxxxxxxxxx
--cram-hmac-alg=sha1
--allow-two-primaries
--ko-count=4
--ping-timeout=5
--ping-int=10
--connect-int=10
--timeout=60
So the max argument count in adm_connect definitely needs to be higher
than 20.
Bye, Martin