NFS mount points causing boot hang when network unavailable

[Steve N.]

Our AIX system has a couple of NFS mount points to a LINUX system that
is used simply as a fax server. During a recent reboot of the system,
we were testing what would happen if we had no connectivity to the
network. To do this, we simply disconnected the cable from the RS/6000
Ethernet card.

What we found was that the system got to the point where it was trying
to mount the NFS areas and simply hung there indefinitely. After about
10-15 minutes, we reconnected the network cable, and the system
proceeded to continue its boot as if it had found what it was waiting
for.

As this is a production system, we want it to proceed to boot up in a
timely manner, regardless if the fax server was reachable or not.

Example of our /etc/filesystems NFS entry:

/apps/fax:
dev = "/apps/fax"
vfs = nfs
nodename = fax_host
mount = true
type = nfs
options = bg,soft,intr
account = false

Our mount points are all set to SOFT mounts, so we thought it would
just bypass these NFS areas if it was not able to reach them. I am
assuming that the reason it hung was because we have not set a value
for these fields:

- NUMBER of times to attempt mount
- NFS TIMEOUT. In tenths of a second

Can anyone confirm this, and if so, what values would you recommend?

Thx,

Steve

[futsnfup]

It might be a good idea to switch the "mount = true" option to false
to prevent your NFS filesystems from mounting automatically at system
restart. Then use the mount command with the -t option for type
(mount -t nfs) later in one of the startup files, like rc.nfs.
JH

[email]steven_nospam@yahoo.ca[/email] (Steve N.) wrote in message news:<a7f2ae7e.0309231037.10bffbd@posting.google.c om>...

> Our AIX system has a couple of NFS mount points to a LINUX system that
> is used simply as a fax server. During a recent reboot of the system,
> we were testing what would happen if we had no connectivity to the
> network. To do this, we simply disconnected the cable from the RS/6000
> Ethernet card.
>
> What we found was that the system got to the point where it was trying
> to mount the NFS areas and simply hung there indefinitely. After about
> 10-15 minutes, we reconnected the network cable, and the system
> proceeded to continue its boot as if it had found what it was waiting
> for.
>
> As this is a production system, we want it to proceed to boot up in a
> timely manner, regardless if the fax server was reachable or not.
>
> Example of our /etc/filesystems NFS entry:
>
> /apps/fax:
> dev = "/apps/fax"
> vfs = nfs
> nodename = fax_host
> mount = true
> type = nfs
> options = bg,soft,intr
> account = false
>
> Our mount points are all set to SOFT mounts, so we thought it would
> just bypass these NFS areas if it was not able to reach them. I am
> assuming that the reason it hung was because we have not set a value
> for these fields:
>
> - NUMBER of times to attempt mount
> - NFS TIMEOUT. In tenths of a second
>
> Can anyone confirm this, and if so, what values would you recommend?
>
> Thx,
>
> Steve