From: Andreas Dilger <[EMAIL PROTECTED]>
Date: Wed, 4 Apr 2007 13:25:11 -0600
On Apr 04, 2007 13:25 -0400, John R. Dunning wrote:
> I'm debugging lustre on my prototype cluster. I frequently see messages
like
>
> Apr 4 13:02:02 sf0-m0n2 [4295282.836000] LustreError:
1193:0:(ldlm_lib.c:700:target_handle_connect()) frost012-OST0002: NID [EMAIL
PROTECTED] (frost012-mdtlov_UUID) reconnected with 1 conn_cnt; cookies not
random?
Are your client nodes diskless, and without any hardware RNG to seed the
/dev/random pool at boot time?
Yes, certainly diskless. I don't know off the top of my head whether there's
anything special planned for what to do about initializing the prng.
Fixing that would be the preferred way
to go, because other services besides Lustre use /dev/[u]random and can
break much more quietly than Lustre.
Try the patch in bug 10802 - that will be going into 1.4.11 and 1.6.1.
Ok, thanks, I'll put it on the list. I probably won't do anything with it
right now, as I'm juggling lots of other things.
_______________________________________________
Lustre-discuss mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
