Re: [PATCH RFC] vfs: make fstatat retry on ESTALE errors from getattrcall

[Scott Lovenberg]

On 4/16/2012 1:46 PM, Jeff Layton wrote:
> NFS will generally return a different error if the process catches a
> fatal signal, so a soft mount should not be necessary and is not
> recommended anyway...
>
> In any case, we loop indefinitely now in the NFS code when (for
> instance) there's a loss of communication. Users are not generally
> happy if that causes an error, since their applications start dying.
From the peanut gallery, I've always set an infinite loop with an 
exponential backoff on the loss of communication.  IE, in some code I 
wrote for S3backer (a FUSE file system on top of Amazon EC3) a few years 
ago (committed by Archie Cobbs).

The trade off is that your applications will try to submit requests if 
you don't tell them "leave me alone, I can't service you now".  The more 
I think about it, the more it seems like failing silently.  Isn't the 
rule supposed to be "if you must fail, do it loudly and as soon as 
possible"?  Just my $0.02.  Take it as you will.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/