[ale] Fstab fails

Mon Aug 29 11:04:15 EDT 2011

On Mon, 2011-08-29 at 10:04 -0400, Wolf Halton wrote: 
> I have been trying to get my nfs thing to work, but it has proven so
> fragile that I am giving up on it.

> So would it make sense to have the configs in a separate place from
> the spot on the local server where they are used, right
> /root/configs          # git repo
> and push them to
> server-00:/configs # home server
> server-01:/configs # 1st remote server
> and so on.
> I don't know how to do it with git and/or puppet.  I have been looking
> at autofs, which is more reminiscent of nfs (but nfs-style mount lines
> fail).  The autofs solution is more aimed at what I think of as
> focus-clusters - 3-server groups that are linked in in their focus -
> for instance, to deliver a specific application suite.

Definitely look closer at autofs.  I use that for almost all NFS work
and, if you're really intent on very dynamic configurations, you can use
network maps or the /net auto directory.  I really use the later pretty
extensively.  You don't even need special configurations to make it work
and you simply open up your shares on your server, which you have to do
anyways!

> I like the "update every five minutes" of puppet.

> Does it make the most sense to have a central puppet server that
> pushes to all the other servers in the network segment, or would it be
> better to have a puppet server on each focus-cluster?

> Wolf

Regards,
Mike

> On Tue, Aug 9, 2011 at 10:20 AM, Avery Ceo <avery.ceo at gmail.com> wrote:
> > None of which is a reason to avoid Puppet.  It is a good way of doing push.
> > Actually, the two can work well together.  Keeping your Puppet configs in a
> > local Git repository is a Good Idea™.
> >
> > On Aug 8, 2011 5:35 PM, "Jim Kinney" <jim.kinney at gmail.com> wrote:
> >> With configs in version control it easy to roll back from a mistake.
> >> Pushing from central repo is required for fast config change : for
> >> servname
> >> in $list do pushconfig&rehup; done
> >> For pull, each must have a key on central or use same key (BAD!). Push
> >> just
> >> needs central key installed by default.
> >> Other reasons to complicated for thumb typing on cell :-)
> >> On Aug 8, 2011 3:49 PM, "Wolf Halton" <wolf.halton at gmail.com> wrote:
> >>> Jim..
> >>>
> >>> I am almost sure that a pull from the clients makes more sense than a
> >> push,
> >>> but as usual I am open to suggestions.
> >>> Why put the config files into a git or svn repo? Why not serve them
> >> directly
> >>> fro the actual files. I have been reading about puppet for a couple of
> >>> hours and sorting out what would be useful here. It seems hugely useful,
> >>> especially if I am pushing out configurations on a larger scale than I
> >>> currently am doing.
> >>>
> >>> I have a test script that uses rsync to pull the files off the server,
> >>> now
> >>> that I am using crontab better, I can use cron to set the script to run
> >> once
> >>> an hour. I could use puppet to call the cron job that runs that on the
> >>> client side, but rsync requires ssh password for the user. I do not
> >>> understand how to set up puppet to pull the file-structure directly, but
> >>> I
> >>> think it would mean I didn't need the rsync script or the cron job, which
> >>> would save a lot of lash-up, and wouldn't have to use an rsync daemon on
> >> the
> >>> server.
> >>>
> >>> -Wolf
> >>>
> >>> On Thu, Aug 4, 2011 at 2:30 PM, Jim Kinney <jim.kinney at gmail.com> wrote:
> >>>
> >>>> I would look at putting the config files in a subversion or git repo and
> >>>> pushing out with repo sync or rsync or using puppet or cfengine to
> >>>> handle
> >>>> this.
> >>>> On Aug 4, 2011 1:49 PM, "Wolf Halton" <wolf.halton at gmail.com> wrote:
> >>>> > Oh, ok.. Thanks,
> >>>> > I have not used automount before. Would you feel safe mounting a
> >> critical
> >>>> > directory with it - not necessarily a remote /home directory?
> >>>> > The nfs share I have up is holding the config files for a web app.
> >>>> > This
> >>>> was
> >>>> > our solution to having to update the config files on several servers
> >>>> > in
> >>>> the
> >>>> > event that they needed to be edited. Without it, my test box has a
> >>>> > copy
> >>>> of
> >>>> > the source on it, but it is plain that this failsafe didn't work all
> >> that
> >>>> > well.
> >>>> >
> >>>> > On Thu, Aug 4, 2011 at 12:57 PM, Lightner, Jeff <JLightner at water.com>
> >>>> wrote:
> >>>> >
> >>>> >> Right – Avery’s email was saying you could use automount rather than
> >>>> >> fstab. With automount the filesystem doesn’t mount at boot time but
> >> does
> >>>> >> mount the first time something tries to use it which should be long
> >>>> after
> >>>> >> boot.****
> >>>> >>
> >>>> >> ** **
> >>>> >>
> >>>> >> Automount also means the filesystem gets unmounted automatically when
> >>>> not
> >>>> >> in use and remounted when used later. One nice side to this is that
> >>>> >> if
> >>>> the
> >>>> >> exporting server goes down and it isn’t currently NFS mounted on your
> >>>> server
> >>>> >> it doesn’t cause problems like hard NFS mounts do. ****
> >>>> >>
> >>>> >> ** **
> >>>> >> ------------------------------
> >>>> >>
> >>>> >> *From:* ale-bounces at ale.org [mailto:ale-bounces at ale.org] *On Behalf
> >>>> >> Of
> >>>> *Wolf
> >>>> >> Halton
> >>>> >> *Sent:* Thursday, August 04, 2011 12:23 PM
> >>>> >> *To:* Atlanta Linux Enthusiasts
> >>>> >>
> >>>> >> *Subject:* Re: [ale] Fstab fails
> >>>> >> ****
> >>>> >>
> >>>> >> ** **
> >>>> >>
> >>>> >> In my case, the system is coming up but just without the nfs
> >>>> >> directory
> >>>> >> mounted. ****
> >>>> >>
> >>>> >> On Aug 4, 2011 9:03 AM, "Lightner, Jeff" <JLightner at water.com> wrote:
> >>>> >> > Good idea.
> >>>> >> >
> >>>> >> >
> >>>> >> >
> >>>> >> >
> >>>> >> >
> >>>> >> >
> >>>> >> > ________________________________
> >>>> >> > From: ale-bounces at ale.org [mailto:ale-bounces at ale.org] On Behalf Of
> >>>> >> Avery Ceo
> >>>> >> > Sent: Wednesday, August 03, 2011 7:42 PM
> >>>> >> > To: Atlanta Linux Enthusiasts; transam at verysecurelinux.com
> >>>> >> > Subject: Re: [ale] Fstab fails
> >>>> >> >
> >>>> >> >
> >>>> >> > Have you considered the automounter? By waiting until the first
> >>>> attempt
> >>>> >> to read a file before mounting the share, you might avoid your
> >>>> >> startup
> >>>> >> issue, and you would get remounts after a lost connection for free if
> >>>> there
> >>>> >> is a network blip.
> >>>> >> > On Aug 3, 2011 6:47 PM, "Bob Toxen" <transam at verysecurelinux.com
> >>>> <mailto:
> >>>> >> transam at verysecurelinux.com>> wrote:
> >>>> >> >> This still will hang in startup (though you'll probably be in
> >>>> >> >> multiuser).
> >>>> >> >>
> >>>> >> >> Specify timeouts in case NFS cannot start, e.g., the server is
> >> down.
> >>>> >> >>
> >>>> >> >> Bob Toxen
> >>>> >> >>
> >>>> >> >> On Wed, Aug 03, 2011 at 04:21:47PM -0400, Andrew Wade wrote:
> >>>> >> >>> With NFS, I set fstab
> >>>> >> >>>
> >>>> >> >>> server:/ifs/fs02 /ifs/fs02 nfs bg,noauto,rsize=8192,wsize=8192 0
> >>>> >> >>> 0
> >>>> >> >>>
> >>>> >> >>> *note the noauto, this tells it to not automount. If it tries to
> >>>> >> automount
> >>>> >> >>> and is unable, your server will stall in startup. So you put this
> >>>> entry
> >>>> >> in
> >>>> >> >>> /etc/rc.local:
> >>>> >> >>>
> >>>> >> >>> mount /ifs/fs02
> >>>> >> >>>
> >>>> >> >>> Also check to make sure that nfs and portmap services are started
> >>>> upon
> >>>> >> boot
> >>>> >> >>> time.
> >>>> >> >>>
> >>>> >> >>> That way, the server starts up and then mounts the nfs under
> >>>> rc.local
> >>>> >> (where
> >>>> >> >>> if the nfs share hangs, it would not affect the normal system
> >>>> startup).
> >>>> >> >>>
> >>>> >> >>>
> >>>> >> >>>
> >>>> >> >>> Andrew
> >>>> >> >>> RHCE
> >>>> >> >>>
> >>>> >> >>> On Tue, Aug 2, 2011 at 5:50 PM, planas
> >>>> >> >>> <jslozier at gmail.com<mailto:
> >>>> >> jslozier at gmail.com>> wrote:
> >>>> >> >>>
> >>>> >> >>> > **
> >>>> >> >>> > Wolf
> >>>> >> >>> >
> >>>> >> >>> > Could configuration be faulty, I asked about setting an
> >>>> >> >>> > internal
> >>>> ntfs
> >>>> >> drive
> >>>> >> >>> > awhile ago and I got this link from Ubuntu:
> >>>> >> >>> >
> >>>> >> >>> > https://help.ubuntu.com/community/Fstab
> >>>> >> >>> >
> >>>> >> >>> > It has links to more network configuration.
> >>>> >> >>> >
> >>>> >> >>> >
> >>>> >> >>> > On Tue, 2011-08-02 at 14:27 -0400, Wolf Halton wrote:
> >>>> >> >>> >
> >>>> >> >>> > Can you think of any reason why fstab would not be read on
> >> reboot.
> >>>> >> Ubuntu
> >>>> >> >>> > lucid have an nfs share that was not mounted automagically when
> >> I
> >>>> >> rebooted
> >>>> >> >>> > the client machine. Broke a bunch of stuff.
> >>>> >> >>> > Wolf
> >>>> >> >>> > PS mount -a picked up and mounted the nfs directory.
> >>>> >> >>> >
> 
> -- 
> This Apt Has Super Cow Powers - http://sourcefreedom.com
> 
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists at
> http://mail.ale.org/mailman/listinfo
> 
> -- 
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
> 
> 

-- 
Michael H. Warfield (AI4NB) | (770) 985-6132 |  mhw at WittsEnd.com
   /\/\|=mhw=|\/\/          | (678) 463-0932 |  http://www.wittsend.com/mhw/
   NIC whois: MHW9          | An optimist believes we live in the best of all
 PGP Key: 0x674627FF        | possible worlds.  A pessimist is sure of it!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 482 bytes
Desc: This is a digitally signed message part
Url : http://mail.ale.org/pipermail/ale/attachments/20110829/38ac5ae5/attachment.bin 