Softupdates On Existing System

Softupdates On Existing System

Post by Tim Daneliu » Wed, 08 Aug 2001 06:00:02



One of the machines here is a production 4.3R box that has a single
disk partition for everything - that is, the whole image (/dev/da0s2a) has only one mount point (/).

How do I get to a running system w/o mounting that partition so that
I can apply the 'tunefs -n enable' to it?
--
------------------------------------------------------------------------------
Tim Daneliuk

 
 
 

Softupdates On Existing System

Post by Michael Sierchi » Wed, 08 Aug 2001 06:04:17



> One of the machines here is a production 4.3R box that has a single
> disk partition for everything - that is, the whole image (/dev/da0s2a) has only one mount point (/).

> How do I get to a running system w/o mounting that partition so that
> I can apply the 'tunefs -n enable' to it?

boot -s
tunefs -n enable /
boot

 
 
 

Softupdates On Existing System

Post by Tim Daneliu » Wed, 08 Aug 2001 06:20:02




> > One of the machines here is a production 4.3R box that has a single
> > disk partition for everything - that is, the whole image (/dev/da0s2a) has only one mount point (/).

> > How do I get to a running system w/o mounting that partition so that
> > I can apply the 'tunefs -n enable' to it?

> boot -s
> tunefs -n enable /
> boot

Tnx - that does the trick nicely...  (Duh!  Why didn't I think if this...
old age ;)

While we're on the subject, what is the *realistic* risk of running soft
updates?  My understanding is that it will enable corruption of the
disk meta-data, but not the file data itself.  Is this correct?

--
------------------------------------------------------------------------------
Tim Daneliuk

 
 
 

Softupdates On Existing System

Post by Mike Andre » Wed, 08 Aug 2001 06:29:59


:>
:> >
:> > One of the machines here is a production 4.3R box that has a single
:> > disk partition for everything - that is, the whole image (/dev/da0s2a) has only one mount point (/).
:> >
:> > How do I get to a running system w/o mounting that partition so that
:> > I can apply the 'tunefs -n enable' to it?
:>
:> boot -s
:> tunefs -n enable /
:> boot

: Tnx - that does the trick nicely...  (Duh!  Why didn't I think if this...
: old age ;)

: While we're on the subject, what is the *realistic* risk of running soft
: updates?  My understanding is that it will enable corruption of the
: disk meta-data, but not the file data itself.  Is this correct?

It is my understanding that softupdates does its disk writes in
such a way that the data and metadata on the disk are always
consistent.

I've been using softupdates since 4.3 came out, and it has not
grommiched any of my disks yet -- despite numerous opportunities
courtesy of my local power failu^wcompany.

--
Basically, no matter what you compare spammers to,
that object will be insulted.
                -- Chris Pickett, in a.s.r

 
 
 

Softupdates On Existing System

Post by Darryl Okahat » Wed, 08 Aug 2001 10:23:24



> While we're on the subject, what is the *realistic* risk of running soft
> updates?  My understanding is that it will enable corruption of the
> disk meta-data, but not the file data itself.  Is this correct?

     I think this has been fixed by now, but an *old* problem with
softupdates is that, IIRC, the system can panic if it runs out of disk
space.  This is more of a problem that it first seems, as softupdates
can take something like up to a minute to free up newly-freed disk
space.  I think the rule-of-thumb was to always insure that you have at
least twice as much maximum disk space free that your system uses up in
one minute.

     However, this was a LONG time ago, and I assume that it's since
been resolved (but I have not seen any mention of this).  Hopefully,
someone with more knowledge will speak up.

--
        Darryl Okahata

DISCLAIMER: this message is the author's personal opinion and does not
constitute the support, opinion, or policy of Agilent Technologies, or
of the little green men that have been following him all day.

 
 
 

Softupdates On Existing System

Post by Steve O'Hara-Smit » Thu, 09 Aug 2001 04:23:32


On 06 Aug 2001 21:20:02 GMT

TD> While we're on the subject, what is the *realistic* risk of running soft
TD> updates?  My understanding is that it will enable corruption of the
TD> disk meta-data, but not the file data itself.  Is this correct?

        Not quite, the disk meta-data is always correct with softupdates
file data will be lost if it is still in unwritten buffers at the time of a
crash. After a crash the only repair a softupdates file system should need
is free space recovery.

        The main problem with softupdates is that free space is not recovered
immediately which is only a problem if you cycle through enough temporary
storage to exhaust the free space faster than it is recovered. Enough free
space to cope with the peak turnover rate is the real (and effective) cure.

--
    Directable Mirrors - A Better Way To Focus The Sun

                        http://www.best.com/~sohara

 
 
 

Softupdates On Existing System

Post by JD » Thu, 09 Aug 2001 11:13:46



> On 06 Aug 2001 21:20:02 GMT

> TD> While we're on the subject, what is the *realistic* risk of running soft
> TD> updates?  My understanding is that it will enable corruption of the
> TD> disk meta-data, but not the file data itself.  Is this correct?

> Not quite, the disk meta-data is always correct with softupdates
> file data will be lost if it is still in unwritten buffers at the time of a
> crash. After a crash the only repair a softupdates file system should need
> is free space recovery.

Another consideration is that writing the real data can be further delayed,
and will potentially be invalid for longer.   However, the risk of a destroyed
filesystem is not increased by the use of softupdates relative to normal
FreeBSD mounts...   In fact, it is probably safer.

John

 
 
 

1. With softupdates, tar brings the system to its knees

I enabled softupdates on the /usr/home slice and on /var. I downloaded
the latest ports tree from freebsd.org and was doing a tar xvzf on the
file. I tried it first in /usr/home/myuserid which has softupdates on
and the system crawled. It took a minute to log on (on another virtual
terminal) and every action was slow (running vi, etc). I tried the
same thing again on /root which does not have softupdates on and
everything was okay, no response time problems on other logins. The
system is modest - 128M memory, 266Mhz CPU. SCSI Wide drive, FreeBSD
4.0.

I thought that softupdates was supposed to speed things up, or at
least not get in the way. This makes me very nervous. Can you imaging
running a tar on your server and the help desk calls start coming in?
Any idea what the problem could be?

2. Installing OSR 5 on Compaq Deskpro 4000/5166

3. X/GNOME problems

4. USB scanner exists to Yast2, does not exist to SANE

5. "Voxware" for FreeBSD v2.1.x

6. rmuser says a user does not exist, but adduser says the user does exist?

7. multiple volumen - auto-spanning archive

8. Mount point does not exist/device fd0 does not exist

9. Move volumes from an existing diskgroup to another existing diskgroup in VxVM

10. Apache 1.3.9: File does not exist: path-to-existing-file

11. How to resize existing file systems / slices ?

12. Installing Linux on an existing MS-DOS system