NFS based /usr prevents normal startup due to slow net init

[Tom Judge]

Steven Hartland wrote:
> Another observation from my recent dealings with using
> NFS based /usr is that the remote critical mounts via
> nfs dont always give the network enough time to
> initialise before running. The first error displayed
> is:
> Mounting NFS file systems:mount_nfs: nfs1: hostname nor servname 
> provided, or not known
> 
> This is particularly noticeable when the machine is
> connected to Cisco equipment as they take quite a
> while link to the connected host after initialisation.
> 
> The result of this is that other services such as
> ldconfig fail to initialise properly due to the
> mount not being available until some point later
> in the boot process once link has been established.
> 
> Has anyone else experienced this?
> 
> Should mountcritremote use "mount -a -t nfs" when this
> appears to return after a short period without said FS's
> being successfully mounted? Is there a way to ensure that
> mount doesnt return without success i.e. a missing flag
> in my fstab or should mountcritremote be updated to test
> for failure and retry?
> 
> This was bought up some time ago on the current
> list but no answers where forth coming:
> http://lists.freebsd.org/pipermail/freebsd-current/2005-December/058935.html 
> 
> 
> Test details:
> Network card: bge0
> Switch: Cisco 6509
> Switch Blade: WS-X6748-GE-TX
> fstab line: nfs1:/fs/usr /usr nfs rw 0 0
> 


On our systems I have modified /etc/rc.d/mountcritremote to have a 15 
second sleep at the beginning to allow for slow port setup time.  Its 
not really a nice fix but it does the job.

Tom