>>>>> "Matt" == Matt Johnston <[EMAIL PROTECTED]> writes:
Matt> On Mon, Mar 28, 2005 at 06:30:28PM -0500, Sam Hartman wrote:
>> >>>>> "Steve" == Steve Langasek <[EMAIL PROTECTED]> writes:
Steve> It seems to me that it would be better to fix this in the
Steve> mount options for the NFS mount in question...
>> Hmm. Actually, will a signal even interrupt an NFS read? It
>> may well be that is the only solution.
Matt> The nfs(5) manpage implies that it will interrupt when
Matt> signalled if mounted with the 'intr' option. I don't believe
Matt> that mounting with 'soft' would be desirable (the mount(8)
Matt> manpage advises against it), so the alarm() would still be
Matt> needed?
I think that applies to the mount not to operations against the mount.
I can see Steve's point here though; we don't want to add nfs-specific
logic to everything. Arguably the filesystem should be responsible
for presenting a usable interface in the case of network problems.
However I can see your point from a practical standpoint.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
