Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 9 Mar 1998 03:01:33 +0000 (GMT)
From:      Terry Lambert <tlambert@primenet.com>
To:        dyson@FreeBSD.ORG
Cc:        karpen@ocean.campus.luth.se, current@FreeBSD.ORG
Subject:   Re: Okay, -current should be conditionally safe to use
Message-ID:  <199803090301.UAA15987@usr08.primenet.com>
In-Reply-To: <199803080012.TAA00282@dyson.iquest.net> from "John S. Dyson" at Mar 7, 98 07:12:31 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> > Would tuning the NFS exported disk SoftUpdates do a significant difference?
> > 
> Yes.
> 
> > For the moment I don't think we're ready to use the SoftUpdates code, since
> > we're using the machine for more then just play, but it might be an
> > interesting thing to try once it's more robust.
>
> It probably isn't a good idea to use softupdates in production yet.  However,
> another interesting thing to try is:
> 
> sysctl -w vfs.nfs.async=1
> 
> on the server.  This is better (safer) than softupdates, but you *can*
> have data lossage, due to writes not being committed to disk.  It is
> a good idea to have a UPS when using the above option.

Note: it is a technical violation of the NFS protocol specification
for an NFS server to ACK a write which has not been commited to
stable storage.

The "Prestoserve" stuff commits the writes to stable storage (battery
backed RAM).  The Auspex and NetWork Appliance stuff works similarly.

The vfs.nfs.async sysctl, Soft Updates, and the SunOS/Solaris/SVR4
write gathering, as well as the SVR4 Delayed Ordered Writes (DOW)
are all, in a sense, technical violations.

For more information, see:

	X/Open CAE Specification C218
	Protocols for X/Open PC Interworking: XNFS, Issue 4
	ISBN: 1-872630-66-9

If you are concerned about fault tolerance, then you should look
into a UPS (#define NFS_SERVER STABLE_STORAGE).


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199803090301.UAA15987>