2003-05-15 17:45:04

by Jeff Layton

[permalink] [raw]
Subject: Re: network storage solutions - mounts

Alvin Oga wrote:

> On Wed, 14 May 2003, Glen Kaukola wrote:
>
> > Alvin Oga wrote:
>
> some kernel folks will tell you that using soft mount WILL cause
> corrupt data ... and that hard mount is better...
>

This is correct. The NFS homepage will explain this pretty
clearly and there's a link to a paper by Chuck Lever at
Netapps that explains it as well:

http://www.netapp.com/tech_library/3183.html

I'm cc-ing the NFS mailing list perhaps they can comment
on this.

> i prefer soft mount ... as i want the other machines to keep going,
> even if one of the other boxes died for some reason
>
> if using hardmounts, everybody sits and wait for the other server
> to come back online... if and when it does, and if it comes up
> at some other ip# or some other nfs ports, other pcs need to be
> rebooted too
> - i dont like waiting around or rebooting
>
> hard mount vs soft mount is another vi vs emacs warz
> but at least there's more technical justifications for the reasons
> ( i've not seen any corrupt data due to soft mounts
> -
> - have seens lots of people that cannot work due to hard mounts
> - onto servers that died
> -
>

I would hazard to say that you've been luckly. However,
the potential is there for corrupt data. I would rather wait
on the server to come back than take a chance on corrupting
data.

> trick is NO rebooting allowed, under any condition ... NOT allowed
> - find another way to fix the problem
> ( well, only the last person in the list/responsibility hits
> ( the button
>
> > On the odd chance my boss does buy some duplicate
> > storage, what would you recommend I use to manage it?
>
> software raid ... on 4 IDE ports ... one drive per cable
>
> for NFS .... i'd use soft mounts
>

I'll let other people from the NFS list comment here. However,
I'd say this is risky and something I would do - especially if
you value your data.

Jeff

>
> and NOBODY gets to reboot/touch any server ...
>

--
Jeff Layton
Senior Engineer - Aerodynamics and CFD
Lockheed-Martin Aeronautical Company - Marietta

"Is it possible to overclock a cattle prod?" - Irv Mullins




-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara
The only event dedicated to issues related to Linux enterprise solutions
http://www.enterpriselinuxforum.com

_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs