ext2 -> ext3 on the fly?

ext2 -> ext3 on the fly?

[Robert L. Harris]

[-- Attachment #1: Type: text/plain, Size: 1269 bytes --]



I have a number of servers which are currently mounting /usr as ext2.  I 
have a means of doing an tune2fs -j on all of them remotely en mass but
I'd rather not reboot them all to enable journaling on machines that are
up and not having issues.  I've tried to do a:

mount -t ext3 -o remount /usr

as well as just a mount -o remount after changing the fstab.

on a test box but it just blows out a usage message.  Is there a way to
do this remount without a complete reboot that'll be transparant to
users?


If not, is it dangerous to tune2fs the filesystems, change the fstab and
then leave the box up for 2-6 months and let them reboot through
atrrition, upgrades, etc?

Current kernel is 2.4.21-ac3, getting outages and upgrades is a rather
long process involving regression testing, etc.

Robert

:wq!
---------------------------------------------------------------------------
Robert L. Harris                     | GPG Key ID: E344DA3B
                                         @ x-hkp://pgp.mit.edu
DISCLAIMER:
      These are MY OPINIONS ALONE.  I speak for no-one else.

Life is not a destination, it's a journey.
  Microsoft produces 15 car pileups on the highway.
    Don't stop traffic to stand and gawk at the tragedy.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: ext2 -> ext3 on the fly?

[David T Hollis]

Robert L. Harris wrote:

>I have a number of servers which are currently mounting /usr as ext2.  I 
>have a means of doing an tune2fs -j on all of them remotely en mass but
>I'd rather not reboot them all to enable journaling on machines that are
>up and not having issues.  I've tried to do a:
>
>mount -t ext3 -o remount /usr
>
>as well as just a mount -o remount after changing the fstab.
>
>on a test box but it just blows out a usage message.  Is there a way to
>do this remount without a complete reboot that'll be transparant to
>users?
>
>
>If not, is it dangerous to tune2fs the filesystems, change the fstab and
>then leave the box up for 2-6 months and let them reboot through
>atrrition, upgrades, etc?
>
>Current kernel is 2.4.21-ac3, getting outages and upgrades is a rather
>long process involving regression testing, etc.
>
>Robert
>
>:wq!
>---------------------------------------------------------------------------
>Robert L. Harris                     | GPG Key ID: E344DA3B
>                                         @ x-hkp://pgp.mit.edu
>DISCLAIMER:
>      These are MY OPINIONS ALONE.  I speak for no-one else.
>
>Life is not a destination, it's a journey.
>  Microsoft produces 15 car pileups on the highway.
>    Don't stop traffic to stand and gawk at the tragedy.
>  
>
I would be inclined to say it's not safe not from a code perspective but 
an administration perspective.  If you make a change as significant as 
the file system format but don't test it until you reboot the box six 
months from now (when you aren't thinking about how you changed it six 
months ago) problems are likely to happen.  Could be a simple typo or 
something else, but it can definitely come back to bite you in the 
backside.  Granted, if you forgot to change your fstab to ext3, you'd 
still boot and be fine running as ext2, you just never can tell what 
could happen.  Murphy always likes to visit on those occasions.

Re: ext2 -> ext3 on the fly?

[Matti Aarnio]

On Fri, Aug 29, 2003 at 10:16:19AM -0400, Robert L. Harris wrote:
> I have a number of servers which are currently mounting /usr as ext2.  I 
> have a means of doing an tune2fs -j on all of them remotely en mass but
> I'd rather not reboot them all to enable journaling on machines that are
> up and not having issues.  I've tried to do a:
> 
> mount -t ext3 -o remount /usr
> as well as just a mount -o remount after changing the fstab.

  No, the 'remount' does not work for this.
  I have done this type of upgrade myself, but I had to unmount
  and then remount filesystems.

> on a test box but it just blows out a usage message.  Is there a way to
> do this remount without a complete reboot that'll be transparant to
> users?

  No.
 
> If not, is it dangerous to tune2fs the filesystems, change the fstab and
> then leave the box up for 2-6 months and let them reboot through
> atrrition, upgrades, etc?

  In one instance I had considerable difficulties to get the box up.
  Or rather it probably came up, but claimed bogously in mtab  having
  been mounted as  ext2,  while it in reality was ext3.

  You might want to run fsck on the filesystems anyway.

> Current kernel is 2.4.21-ac3, getting outages and upgrades is a rather
> long process involving regression testing, etc.

  Invisible upgrade isn't possible, unfortunately.

> Robert
> ---------------------------------------------------------------------------
> Robert L. Harris                     | GPG Key ID: E344DA3B

/Matti Aarnio

Re: ext2 -> ext3 on the fly?

[Robert L. Harris]

[-- Attachment #1: Type: text/plain, Size: 2848 bytes --]

Thus spake David T Hollis (dhollis@davehollis.com):

> Robert L. Harris wrote:
> 
> >I have a number of servers which are currently mounting /usr as ext2.  I 
> >have a means of doing an tune2fs -j on all of them remotely en mass but
> >I'd rather not reboot them all to enable journaling on machines that are
> >up and not having issues.  I've tried to do a:
> >
> >mount -t ext3 -o remount /usr
> >
> >as well as just a mount -o remount after changing the fstab.
> >
> >on a test box but it just blows out a usage message.  Is there a way to
> >do this remount without a complete reboot that'll be transparant to
> >users?
> >
> >
> >If not, is it dangerous to tune2fs the filesystems, change the fstab and
> >then leave the box up for 2-6 months and let them reboot through
> >atrrition, upgrades, etc?
> >
> >Current kernel is 2.4.21-ac3, getting outages and upgrades is a rather
> >long process involving regression testing, etc.
> >
> >Robert
> >
> >:wq!
> >---------------------------------------------------------------------------
> >Robert L. Harris                     | GPG Key ID: E344DA3B
> >                                        @ x-hkp://pgp.mit.edu
> >DISCLAIMER:
> >     These are MY OPINIONS ALONE.  I speak for no-one else.
> >
> >Life is not a destination, it's a journey.
> > Microsoft produces 15 car pileups on the highway.
> >   Don't stop traffic to stand and gawk at the tragedy.
> > 
> >
> I would be inclined to say it's not safe not from a code perspective but 
> an administration perspective.  If you make a change as significant as 
> the file system format but don't test it until you reboot the box six 
> months from now (when you aren't thinking about how you changed it six 
> months ago) problems are likely to happen.  Could be a simple typo or 
> something else, but it can definitely come back to bite you in the 
> backside.  Granted, if you forgot to change your fstab to ext3, you'd 
> still boot and be fine running as ext2, you just never can tell what 
> could happen.  Murphy always likes to visit on those occasions.


My thoughts exactly, just wanted someone else to say it so I can give it
to management.  Unfortunately I know their response will be that our
package management system will ensure this won't happen which
technically it shouldn't but I can think of a couple ways ti still
could.


:wq!
---------------------------------------------------------------------------
Robert L. Harris                     | GPG Key ID: E344DA3B
                                         @ x-hkp://pgp.mit.edu
DISCLAIMER:
      These are MY OPINIONS ALONE.  I speak for no-one else.

Life is not a destination, it's a journey.
  Microsoft produces 15 car pileups on the highway.
    Don't stop traffic to stand and gawk at the tragedy.

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: ext2 -> ext3 on the fly?

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 479 bytes --]

On Fri, 29 Aug 2003 10:27:22 EDT, "Robert L. Harris" said:

> My thoughts exactly, just wanted someone else to say it so I can give it
> to management.  Unfortunately I know their response will be that our
> package management system will ensure this won't happen which
> technically it shouldn't but I can think of a couple ways ti still
> could.

Quite (possibly unintentionally) amusing, because...

> :wq!

Getting one of these in the file is one of the ways, probably... :)

[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]