Strange TCP-related hang on startup w/ recent CVSUP

[Kris Kennaway]

On Wed, Jun 08, 2005 at 12:37:59AM -0500, Terry Kennedy wrote:
>   I have a number of boxes which have been running well w/ 5.4-stable. The 
> last kernel and userland builds on those boxes were on May 26th, from a 
> CVSUP done that same day.
> 
>   Last night I did another CVSUP and the usual buildworld/buildkernel/in-
> stallkernel/installworld/mergemaster, and the boxes wouldn't come back up,
> hanging in sbwait state at the point where they would normally do their
> first net access.
> 
>   In one box, this is a batch of NFS mounts; on the other it is the initial
> ntpdate query. At this point the boxes don't respond to pings and sit for-
> ever (at least an hour) not doing anything. Typing ^C on the console aborts
> the hanging process, and then the startup proceeds. With a ping going from
> another system, I see that the net doesn't come up on the problem boxes until
> several seconds into the execution of the next network-related command - it
> is almost as if the previous ifconfig (done from rc.conf) didn't have any
> effect (despite the "link up" message on the console). The other oddity is
> that I get a "rpc.lockd: 100024 RPC: Port mapper failure" right after lockd
> starts, even though other net commands have completed successfully.
> 
>   Booting from kernel.old (the May 26th one) boots normally, so it is a ker-
> nel issue, not something in userland.

Are you sure you didn't change your kernel config or forget to rebuild
modules?

Kris
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20050608/9bc4fdd6/attachment.bin