ts-7000
[Top] [All Lists]

[ts-7000] Re: strange buffering of nfs file

To:
Subject: [ts-7000] Re: strange buffering of nfs file
From: "j.chitte" <>
Date: Tue, 16 Jun 2009 16:17:23 -0000
--- In  "j.chitte" <> wrote:
>
> --- In  "j.chitte" <j.chitte@> wrote:
> >
> > --- In  "j.chitte" <j.chitte@> wrote:
> > >
> > > I'll try to stick to editing directly on the host machine. If the nfs 
> > > host is allowed to lie about the state of a file I'm not sure running 
> > > sync on the board will affect much. I'll try next time it happends.
> > > 
> > > thx.
> > >
> > 
> > No, this is a big problem.
> > 
> > Even an edit on the host and sync on both systems is leaving files out of 
> > sync.
> > 
> > I'm now editing lighttpd.conf and it's not updating.
> > 
> > I uncommented a module line in the conf and saved it but it's not showing 
> > up on the board.
> > 
> > To verify it was not finger trouble I ran touch on the file one the host 
> > and did ls -ail on the SBC. 
> > 
> > I see the timestamp is updated but the contents of the file remain un 
> > edited when viewed on the ARM. 
> > 
> > It seems the directory is updating but not the file.
> > 
> > What's more worrying is that this is not systematic. I have done a fair bit 
> > of editing on this file in the past hours without problems.
> > 
> > Now it's failing to be consistant again.
> > 
> > :?
> >
> 
> Urrgh. This gets more ugly the closer I look.
> 
> I token change I did on the host got reverted. 
> 
> Even after rebooting the SBC and restarting nfs server on the host I have two 
> different versions depending upon which end I view it from.
> 
> This is MESSY.
> 
> I knew nfs had some issues but this is actually unusable. This file refuses 
> to change.
> 
> I repeated an newline inserted into the file using nano on the host. This is 
> visible in the length and timestamp of the file. If I view the file from from 
> the SBC , I see the new date and length but if I open it with vi the break of 
> the newline is not there!
> 
> If I then save it :wq without even entering an edit mode the new line is gone 
> viewed from either end.
> 
> 
> It gets worse! 
> 
> I remove a hash to uncomment a line in vi. I save and view the file length. 
> It's one less, fine. I reopen with vi and the hash is back in. I've lost my 
> change. I repeat the edit , save and exit.
> 
> One less again! The file content is the same but it's getting smaller !!!
> 
> There is BAD BORK in the air. The directory entries are no long matching the 
> file contents. 
> 
> :/
>

OK after considerable wasted time, this turns out to be a known bug in nfs4. 
How something this flakey gets into the mainline kernel even as "experimental" 
is a bit worrying. 

Anyone playing with a recent kernel should be aware that it is badly borked and 
stick with nfs3. 






------------------------------------

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/ts-7000/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/ts-7000/join
    (Yahoo! ID required)

<*> To change settings via email:
     
    

<*> To unsubscribe from this group, send an email to:
    

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

<Prev in Thread] Current Thread [Next in Thread>
Admin

Disclaimer: Neither Andrew Taylor nor the University of NSW School of Computer and Engineering take any responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the birding-aus mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU