Re: virtual machines losing touch with LAN on reboot?

2008-05-21 Thread Amos Shapira
On Wed, May 21, 2008 at 8:49 PM, Ira Abramov <[EMAIL PROTECTED]>
wrote:

> one of my clients started playing around with vitual machine images on
> VMware player, VMware server (free edition) and Virtualbox. we tried
> hosting ready-made images from the net as well as local installs of
> Ubuntu 7.10, 8.04 and Fedora8 with an XPO machine as host. in all cases


I'm not sure how much this is related but I see something similar on our
setup.

Our setup is CentOS 5 running both on Dom0 and DomU. The DomU's have at
least two separate IP addresses assigned to each one of them (statically for
now, until I get to move to virtual rack and VIP's). The machine setup
process destroys the xen image and re-builds it. In the process it reboots
the Xen guest a couple of times.

I see on one of the Xen servers that sometimes (after too many DomU
reboots?) the second IP address on the re-created DomU just stops
responding. I can access it from Dom0, and I can even see packets from the
outside world reach the Xen Dom0, but can't talk to that second address from
outside. The first IP address works fine.

Have you tried to reboot the VMware host?

--Amos


Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Ehud Karni
On Wed, 21 May 2008 16:43:23 Ira Abramov wrote:
>
> > I can send you a script we use on our system but most of it is our
> > specific details.
>
> you are not scaring me, sounds delicious :-)
> Thanks in advance!

Below is the script (with minor changes). You should also look at the
"rsnapshot" - http://www.rsnapshot.org/ which does the same (with much
more options) in perl.

Ehud


/bin/sh -ex
# Compile by: /bin/sh -ex $*
# --

get_size ()# size left on $1 in MB
{  # for deleting snapshots
set +x
df -m $1 | tail -n 1 | sed -e "s/ [ ]*/ /g" | cut "-d " -f4
set -x
}
# --

HOUR=`date +%H`# hour, HH (10, 13, 16, 22)
case "$HOUR" in# check hour (22/23 or else)
   2? )  MAXPRC=20;;   # No. of rsync processes
   * )   MAXPRC=4 ;;   # Only 4 processes of rsync at 
noon
esac

DIRS="/fs-gib/snapshots"   # snapshots are here
SNAP=`date +%Y%m%d_%H%M`   # snapshot directory name
PRV=`ls -1atd $DIRS/2*_ | head -n 1`   # latest snapshot (update source)

ORG="/snap-$SNAP"  # source (lvm2 snapshot) name
DST="$DIRS/.NXT"   # target (temp name)
DSNP="$DIRS/$SNAP" # target (final name)

lvcreate -v -L60G -pr -s -n $SNAP ulvm/nfs # creates /dev/ulvm/$SNAP

mkdir -p $ORG  # mount point for snapshot
##  mount -t xfs -o ro,nouuid /dev/ulvm/$SNAP $ORG# mount it read-only, 
nouuid check (for xfs)
mount -t ext3 -o ro /dev/ulvm/$SNAP $ORG   # mount it read-only

if [ ! -d $DST ] ; then# copy of last saved snapshot
df -m /fs-gib  # NOT exist, show size & copy
cp -al $PRV $DST   # copy latest snap by Hard links
chmod 700 $DST # ONLY root has access to the 
snapshot
fi

touch -r / $DST# Set time of snapshot (to last 
change of /)

##
## rsync-snapshot.sh is our rsync enhanced script   ##
##  ##
## you should use something like:   ##
##OPTIONS="--verbose --archive --delete --whole-file"   ##
##rsync $OPTIONS "$ORG" "$DST" > "$LIST" 2>&1   ##
##
rsync-snapshot.sh $ORG $DST $MAXPRC ON IGN # update snapshot by rsyncing 
(keep lst files)

touch -t ${SNAP:0:8}${SNAP:9:4} $DST $DST/bru/SNAP-DATE# Set time of 
snapshot (to creation time)

sleep 10

umount $ORG# unmount it
rmdir  $ORG# remove temp dir
lvdisplay
lvremove -f ulvm/$SNAP # remove snapshot

mv $DST $DSNP  # rename to destination name
chmod 755 $DSNP# allow normal users to see it

cp -al $DSNP $DST  # copy (by hard linking) to .NXT
touch -t 200102030405 $DST # make it somewhere in the past
chmod 700 $DST # .NXT should be non reachable

df -m $DST # show disk statistics
SIZE=`get_size $DST`   # get size left on /ldsk

while [ $SIZE -lt 25000 ]  # less than 15 GB
do
LAST=`ls -a1td $DIRS/20* | tail -n 1`  # oldest snapshot
rm -rf $LAST
df -m $DST # show after snapshot delete
SIZE=`get_size $DST`
done

df -m  # show all mounts again

## end of create-snapshot.job 
##


--
 Ehud Karni   Tel: +972-3-7966-561  /"\
 Mivtach - Simon  Fax: +972-3-7966-667  \ /  ASCII Ribbon Campaign
 Insurance agencies   (USA) voice mail and   X   Against   HTML   Mail
 http://www.mvs.co.il  FAX:  1-815-5509341  / \
 GnuPG: 98EA398D Better Safe Than Sorry

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Ira Abramov
Quoting Ehud Karni, from the post of Wed, 21 May:
> > (which zumastor is built to solve). people here wanted hourly0,1,2 and
> > daily0,1,2 AND weekly0,1,2. that means 9 growing snapshots at any given
> > moment, and that is one very serious disk handicap. 
> 
> You don't want many snapshots because of the heavy load penalty (each
> write is multiplied by the number of snapshots).

yup, that's what I said :)

> I can send you a script we use on our system but most of it is our
> specific details.

you are not scaring me, sounds delicious :-)
Thanks in advance!

-- 
The seventh seal
Ira Abramov
http://ira.abramov.org/email/

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Ehud Karni
On Wed, 21 May 2008 16:13:16 +0300, Erez D <[EMAIL PROTECTED]> wrote:
>
> On Wed, May 21, 2008 at 4:06 PM, Ehud Karni <[EMAIL PROTECTED]> wrote:
>
> > On Wed, 21 May 2008 15:15:51 +0300, Erez D <[EMAIL PROTECTED]> wrote:
> > > >
> > > > If you have enough disk space you can create "pseudo" snapshots by
> > > > using hardlinks between the snapshots so only changed file takes space.
> > > > It really depends on the file size distribution - few large files that
> > > > change constantly will defeat this concept, while many small files is
> > > > ideal for this scenario.
> > >
> > > using hard links as snapshot is works only if you ensure people do not
> > open
> > > new files with append mode.
> > > if they do, they will modify your file as well as it's snapshot ...
> >
> > This is not the case with my suggestion. The hard link is between
> > "snapshots", there is no hard link to the original files.
>
>
> you mean you utilize something like rsnapshot ?
>
> this is a great solution. however it is not a real snapshot solution like
> LVM (version whatever)
> this is neither 'create snapshot in zero time' nor 'in zero space', contrary
> to LVM.
>
> total disk space is at least twice the size of your data
> in LVM is only once + size of changes.

It is exactly like rsnapshot (pointed out by Didi).

Ehud.


--
 Ehud Karni   Tel: +972-3-7966-561  /"\
 Mivtach - Simon  Fax: +972-3-7966-667  \ /  ASCII Ribbon Campaign
 Insurance agencies   (USA) voice mail and   X   Against   HTML   Mail
 http://www.mvs.co.il  FAX:  1-815-5509341  / \
 GnuPG: 98EA398D Better Safe Than Sorry

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Erez D
On Wed, May 21, 2008 at 4:06 PM, Ehud Karni <[EMAIL PROTECTED]> wrote:

> On Wed, 21 May 2008 15:15:51 +0300, Erez D <[EMAIL PROTECTED]> wrote:
> > >
> > > If you have enough disk space you can create "pseudo" snapshots by
> > > using hardlinks between the snapshots so only changed file takes space.
> > > It really depends on the file size distribution - few large files that
> > > change constantly will defeat this concept, while many small files is
> > > ideal for this scenario.
> >
> > using hard links as snapshot is works only if you ensure people do not
> open
> > new files with append mode.
> > if they do, they will modify your file as well as it's snapshot ...
>
> This is not the case with my suggestion. The hard link is between
> "snapshots", there is no hard link to the original files.


you mean you utilize something like rsnapshot ?

this is a great solution. however it is not a real snapshot solution like
LVM (version whatever)
this is neither 'create snapshot in zero time' nor 'in zero space', contrary
to LVM.

total disk space is at least twice the size of your data
in LVM is only once + size of changes.



>
> I forgot to add (it seems trivial) that the pseudo "snapshots" are
> mounted read-only.
>
> Ehud.
>
>
> --
>  Ehud Karni   Tel: +972-3-7966-561  /"\
>  Mivtach - Simon  Fax: +972-3-7966-667  \ /  ASCII Ribbon Campaign
>  Insurance agencies   (USA) voice mail and   X   Against   HTML   Mail
>  http://www.mvs.co.il  FAX:  1-815-5509341  / \
>  GnuPG: 98EA398D Better Safe Than Sorry
>


Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Ehud Karni
On Wed, 21 May 2008 15:15:51 +0300, Erez D <[EMAIL PROTECTED]> wrote:
> >
> > If you have enough disk space you can create "pseudo" snapshots by
> > using hardlinks between the snapshots so only changed file takes space.
> > It really depends on the file size distribution - few large files that
> > change constantly will defeat this concept, while many small files is
> > ideal for this scenario.
>
> using hard links as snapshot is works only if you ensure people do not open
> new files with append mode.
> if they do, they will modify your file as well as it's snapshot ...

This is not the case with my suggestion. The hard link is between
"snapshots", there is no hard link to the original files.

I forgot to add (it seems trivial) that the pseudo "snapshots" are
mounted read-only.

Ehud.


--
 Ehud Karni   Tel: +972-3-7966-561  /"\
 Mivtach - Simon  Fax: +972-3-7966-667  \ /  ASCII Ribbon Campaign
 Insurance agencies   (USA) voice mail and   X   Against   HTML   Mail
 http://www.mvs.co.il  FAX:  1-815-5509341  / \
 GnuPG: 98EA398D Better Safe Than Sorry

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Erez D
On Wed, May 21, 2008 at 2:41 PM, Ehud Karni <[EMAIL PROTECTED]> wrote:

> On Wed, 21 May 2008 12:24:01 +0300, Ira Abramov wrote:
> >
> > problem is that the CAD tools running here dictate that we stick to
> > RHEL4 (damn you, Cadence! how hard is it to support RHEL5?!), and
> > snapshots are problematic at 2.6.9 (some problems were solved only at
> > 2.6.16)
> >
> > > > The more I read about ZumaStor I like it more, but the big boon in
> LVM2
> > > > is that I don't have to recompile RHEL's kernel. so any
> production-env
> > > > experiance shared will be appreciated.
> > >
> > > That's a big advantage if you ask me.
> > > What's a killer argument against LVM2?
> >
> > instability in the old kernels, multiple snapshots mean multiple COWs
> > (which zumastor is built to solve). people here wanted hourly0,1,2 and
> > daily0,1,2 AND weekly0,1,2. that means 9 growing snapshots at any given
> > moment, and that is one very serious disk handicap. snapshots in LVM2 are
> > not really designed for that, they were mainly meant to be for hot
> > backups of file systems, so you can freeze the FS just long enough to
> > snap the state to a tarball and remove the snapshot.
>
> You don't want many snapshots because of the heavy load penalty (each
> write is multiplied by the number of snapshots).
>
> If you have enough disk space you can create "pseudo" snapshots by
> using hardlinks between the snapshots so only changed file takes space.
> It really depends on the file size distribution - few large files that
> change constantly will defeat this concept, while many small files is
> ideal for this scenario.



using hard links as snapshot is works only if you ensure people do not open
new files with append mode.
if they do, they will modify your file as well as it's snapshot ...


>
>
> I use this approach and keep more than 60 "snapshots" of 350 GB disk
> on a 700 GB disk.
>
> The whole process is very simple:
>Initiation: copy the whole disk (or an LVM snapshot of it) to the
>destination disk in a SUBDIRECTORY (eg `cp -a').
>
>  For each snapshot:
>1. Check the size left on the "snapshot" disk and remove old
>   "snapshot"s until you have predefined minimum space.
>2. Copy the subdir (hardlink only) to a new subdir (I use `cp -al').
>3. Create an LVM snapshot of the original system.
>4. Sync the copied subdir with the real LVM snapshot (`rsync').
>5. Remove the LVM snapshot.
>
> I can send you a script we use on our system but most of it is our
> specific details.
>
> Ehud
>
>
> --
>  Ehud Karni   Tel: +972-3-7966-561  /"\
>  Mivtach - Simon  Fax: +972-3-7966-667  \ /  ASCII Ribbon Campaign
>  Insurance agencies   (USA) voice mail and   X   Against   HTML   Mail
>  http://www.mvs.co.il  FAX:  1-815-5509341  / \
>  GnuPG: 98EA398D Better Safe Than Sorry
>
> =
> To unsubscribe, send mail to [EMAIL PROTECTED] with
> the word "unsubscribe" in the message body, e.g., run the command
> echo unsubscribe | mail [EMAIL PROTECTED]
>
>


Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Yedidyah Bar-David
On Wed, May 21, 2008 at 02:41:59PM +0300, Ehud Karni wrote:
> If you have enough disk space you can create "pseudo" snapshots by
> using hardlinks between the snapshots so only changed file takes space.
> It really depends on the file size distribution - few large files that
> change constantly will defeat this concept, while many small files is
> ideal for this scenario.
> 
> I use this approach and keep more than 60 "snapshots" of 350 GB disk
> on a 700 GB disk.
> 
> The whole process is very simple:
> Initiation: copy the whole disk (or an LVM snapshot of it) to the
> destination disk in a SUBDIRECTORY (eg `cp -a').
> 
>   For each snapshot:
> 1. Check the size left on the "snapshot" disk and remove old
>"snapshot"s until you have predefined minimum space.
> 2. Copy the subdir (hardlink only) to a new subdir (I use `cp -al').
> 3. Create an LVM snapshot of the original system.
> 4. Sync the copied subdir with the real LVM snapshot (`rsync').
> 5. Remove the LVM snapshot.
> 
> I can send you a script we use on our system but most of it is our
> specific details.

There are also "packaged" scripts that do something like this. I
personally use rsnapshot, there are a few others.
-- 
Didi


=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Ehud Karni
On Wed, 21 May 2008 12:24:01 +0300, Ira Abramov wrote:
>
> problem is that the CAD tools running here dictate that we stick to
> RHEL4 (damn you, Cadence! how hard is it to support RHEL5?!), and
> snapshots are problematic at 2.6.9 (some problems were solved only at
> 2.6.16)
>
> > > The more I read about ZumaStor I like it more, but the big boon in LVM2
> > > is that I don't have to recompile RHEL's kernel. so any production-env
> > > experiance shared will be appreciated.
> >
> > That's a big advantage if you ask me.
> > What's a killer argument against LVM2?
>
> instability in the old kernels, multiple snapshots mean multiple COWs
> (which zumastor is built to solve). people here wanted hourly0,1,2 and
> daily0,1,2 AND weekly0,1,2. that means 9 growing snapshots at any given
> moment, and that is one very serious disk handicap. snapshots in LVM2 are
> not really designed for that, they were mainly meant to be for hot
> backups of file systems, so you can freeze the FS just long enough to
> snap the state to a tarball and remove the snapshot.

You don't want many snapshots because of the heavy load penalty (each
write is multiplied by the number of snapshots).

If you have enough disk space you can create "pseudo" snapshots by
using hardlinks between the snapshots so only changed file takes space.
It really depends on the file size distribution - few large files that
change constantly will defeat this concept, while many small files is
ideal for this scenario.

I use this approach and keep more than 60 "snapshots" of 350 GB disk
on a 700 GB disk.

The whole process is very simple:
Initiation: copy the whole disk (or an LVM snapshot of it) to the
destination disk in a SUBDIRECTORY (eg `cp -a').

  For each snapshot:
1. Check the size left on the "snapshot" disk and remove old
   "snapshot"s until you have predefined minimum space.
2. Copy the subdir (hardlink only) to a new subdir (I use `cp -al').
3. Create an LVM snapshot of the original system.
4. Sync the copied subdir with the real LVM snapshot (`rsync').
5. Remove the LVM snapshot.

I can send you a script we use on our system but most of it is our
specific details.

Ehud


--
 Ehud Karni   Tel: +972-3-7966-561  /"\
 Mivtach - Simon  Fax: +972-3-7966-667  \ /  ASCII Ribbon Campaign
 Insurance agencies   (USA) voice mail and   X   Against   HTML   Mail
 http://www.mvs.co.il  FAX:  1-815-5509341  / \
 GnuPG: 98EA398D Better Safe Than Sorry

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: Listening 103fm streaming

2008-05-21 Thread Cyril SCETBON

Actually, it works like a charm.

Thanks

Hetz Ben Hamo wrote:

Hi,

Just checked. the URL should be: mms://s18wm.castup.net/995460001-52.wmv

It works with mplayer.

Hetz

On Mon, May 19, 2008 at 11:48 AM, Cyril SCETBON <[EMAIL PROTECTED]> wrote:

Hi,

the links mms://live.103.fm/103fm-high/ and mms://live.103.fm/103fm-low do
not respond anymore. I've visited the website and found that I can listen
the radio using
mms://s18wm.castup.net/995460001-52.wmv?ct=FR&rg=WE&aid=546&st=0&ts=0&cu=25E00368-62FB-4A01-A898-9876ED9942CF
, but I don't if this is a temporary link (seems to be) or not .

Has anyone more information on how to catch the stream ?
They don't use GeoIP to authorize only Israeli people ?

Thanks

--
Cyril SCETBON


=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]








--
Cyril SCETBON


=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: virtual machines losing touch with LAN on reboot?

2008-05-21 Thread Hetz Ben Hamo
I think there's a patch for the vmware tools to fix this issue,
although I don't remember where. I also think that it was fixed in
VMWare server 2 beta, if you want to play with.

Thanks,
Hetz

On Wed, May 21, 2008 at 12:49 PM, Ira Abramov
<[EMAIL PROTECTED]> wrote:
> one of my clients started playing around with vitual machine images on
> VMware player, VMware server (free edition) and Virtualbox. we tried
> hosting ready-made images from the net as well as local installs of
> Ubuntu 7.10, 8.04 and Fedora8 with an XPO machine as host. in all cases
> we add a few packages (not upgrading kernel), reboot and find that we
> have no real LAN connection. ethtool shows link is up but no DHCP
> replies are coming. I bumped into Network-Manager there and could not
> figure out where it stores the configs so on the Ubuntu I ust removed it
> (and saw the eth definition show up in /etc/network/interfaces instead,
> yay!), but a reboot left me still with no LAN connection.
>
> I thought it was the VMware tools I compiled, but the same happend on
> Virtualbox! and both before and after the reboot, it was the same
> pcnet32 driver, taking it out and insmodding again made no difference.
>
> then I accidentaly discovered booting at Single-luser mode left me with
> working LAN and dhcp, which worked if I then continued to runlevel 2
> (Ubuntu), but booting to runlevel2 the normal way left me disconnected
> again. this is REPEATABLE.
>
> till I find a solution, I told the guy to boot to "recovery mode" and
> just hit "ctrl-D" at the single-user-mode prompt, and then LAN works.
>
> anyone got an explanation for this extremely odd behvior?!
>
> --
> The Smartest Man on Earth
> Ira Abramov
> http://ira.abramov.org/email/
>
> =
> To unsubscribe, send mail to [EMAIL PROTECTED] with
> the word "unsubscribe" in the message body, e.g., run the command
> echo unsubscribe | mail [EMAIL PROTECTED]
>
>



-- 
Skepticism is the lazy person's default position.
my blog (hebrew): http://benhamo.org

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



virtual machines losing touch with LAN on reboot?

2008-05-21 Thread Ira Abramov
one of my clients started playing around with vitual machine images on
VMware player, VMware server (free edition) and Virtualbox. we tried
hosting ready-made images from the net as well as local installs of
Ubuntu 7.10, 8.04 and Fedora8 with an XPO machine as host. in all cases
we add a few packages (not upgrading kernel), reboot and find that we
have no real LAN connection. ethtool shows link is up but no DHCP
replies are coming. I bumped into Network-Manager there and could not
figure out where it stores the configs so on the Ubuntu I ust removed it
(and saw the eth definition show up in /etc/network/interfaces instead,
yay!), but a reboot left me still with no LAN connection.

I thought it was the VMware tools I compiled, but the same happend on
Virtualbox! and both before and after the reboot, it was the same
pcnet32 driver, taking it out and insmodding again made no difference.

then I accidentaly discovered booting at Single-luser mode left me with
working LAN and dhcp, which worked if I then continued to runlevel 2
(Ubuntu), but booting to runlevel2 the normal way left me disconnected
again. this is REPEATABLE.

till I find a solution, I told the guy to boot to "recovery mode" and
just hit "ctrl-D" at the single-user-mode prompt, and then LAN works.

anyone got an explanation for this extremely odd behvior?!

-- 
The Smartest Man on Earth
Ira Abramov
http://ira.abramov.org/email/

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: LVM2 and other snapshots to emulate Netapp Filer's?

2008-05-21 Thread Ira Abramov
Quoting Amos Shapira, from the post of Wed, 21 May:
> > > LVM2 snapshots and a wrapper script?
> 
> 
> We use LVM2 snapshots for quick xen guest rebuilds (snapshot a base image,
> run automatic script to tweak for particular host instance). Works well for
> a few months now.

problem is that the CAD tools running here dictate that we stick to
RHEL4 (damn you, Cadence! how hard is it to support RHEL5?!), and
snapshots are problematic at 2.6.9 (some problems were solved only at
2.6.16)

> > The more I read about ZumaStor I like it more, but the big boon in LVM2
> > is that I don't have to recompile RHEL's kernel. so any production-env
> > experiance shared will be appreciated.
> 
> That's a big advantage if you ask me.
> What's a killer argument against LVM2?

instability in the old kernels, multiple snapshots mean multiple COWs
(which zumastor is built to solve). people here wanted hourly0,1,2 and
daily0,1,2 AND weekly0,1,2. that means 9 growing snapshots at any given
moment, and that is one very serious disk handicap. snapshots in LVM2 are
not really designed for that, they were mainly meant to be for hot
backups of file systems, so you can freeze the FS just long enough to
snap the state to a tarball and remove the snapshot.

-- 
Your psychic friend
Ira Abramov
http://ira.abramov.org/email/

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: controlling CPU's fan speed

2008-05-21 Thread Boaz Rymland
Hi,

Thanks for the reply.
I didn't mentioned it verbosely but its a desktop machine, not a laptop.
I'm not sure it means too much but I can add that when powering up the
machine I do hear the FAN running at high speed but when it gets to OS
loading phase it gets calmed down - for good until the next reboot. 
I would not want to cycle down the CPU in order to lower the heat. Not
that I think it'll help: when the machine gets loaded (and even some
sites when surfing via firefox cause lots of load) it gets warm very
quickly (so cycling CPU down in advance wont "gather coldness" for
tougher times. I think).

Boaz.

> On my old HP nc8000 the fan speed was inaccessible through the OS. It would
> power up (and make a lot of noise) only on really high loads and the laptop
> would still get pretty hot. I had to use the powersaved daemon to
> (automatically) throttle the CPU freq when not needed.
> 
> AFAIK this is the correct way to keep the temp down on laptops. Temps > 50c
> are normal on high load situations. My current Dell (xps m1530) could boil
> an egg when running 100%.
> 
> One more thing. After 3 years of continuous usage, my nc8000 cpu's airflow
> got blocked by a big solid pile of dirt. This resulted in extremely high
> temps (even the keyboard was burning). Don't forget to check yours for dust
> and dirt!
> 
> Alex
> 
> On Tue, May 20, 2008 at 9:25 PM, Boaz Rymland <[EMAIL PROTECTED]> wrote:
> 
> >  Hi,
> >
> >
> >  I've recently discovered that on my HP Pavillion 6010a the CPU fan is
> > most likely always running at its lowest CPU fan speed. This results in
> > relatively high CPU temperatures - when machine is loaded (and since its
> > running Gentoo its loaded quite a few times ;-), its reaching 70 degrees
> > Celsius. I know this still might be in the green zone for the AMD Athlon 64
> > (single core) its using but I'd rather keep it in even greener area of its
> > threshold.
> >
> > I've searched the net and played with some stuff but mostly didn't find
> > exactly the utility I'm looking for. I can use cpufrequtils to lower CPU
> > cycles, which works ok, but why would I prefer hampering my machine's
> > performance instead of simply pushing the fan cycles up?... .
> >
> > I took a look at lm-sensors but it has a too big config file
> > (/etc/sensors.conf) for my time constraints right now and I'm not sure it'll
> > help.
> >
> > In /proc/acpi/fan/FAN I have only "state" which when "cat"ing, prints:
> > "status: on".
> >
> >
> >  Anyone knows how to hasten the CPU speed on such a system?
> >
> > TIA,
> >
> > Boaz.
> >
> >
> >
> 
> 
> -- 
> |
> | Alex Alexander
> | http://linuxized.blogspot.com
> | http://www.nerd.gr
> \

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]



Re: controlling CPU's fan speed

2008-05-21 Thread Alex Alexander
Hey

On my old HP nc8000 the fan speed was inaccessible through the OS. It would
power up (and make a lot of noise) only on really high loads and the laptop
would still get pretty hot. I had to use the powersaved daemon to
(automatically) throttle the CPU freq when not needed.

AFAIK this is the correct way to keep the temp down on laptops. Temps > 50c
are normal on high load situations. My current Dell (xps m1530) could boil
an egg when running 100%.

One more thing. After 3 years of continuous usage, my nc8000 cpu's airflow
got blocked by a big solid pile of dirt. This resulted in extremely high
temps (even the keyboard was burning). Don't forget to check yours for dust
and dirt!

Alex

On Tue, May 20, 2008 at 9:25 PM, Boaz Rymland <[EMAIL PROTECTED]> wrote:

>  Hi,
>
>
>  I've recently discovered that on my HP Pavillion 6010a the CPU fan is
> most likely always running at its lowest CPU fan speed. This results in
> relatively high CPU temperatures - when machine is loaded (and since its
> running Gentoo its loaded quite a few times ;-), its reaching 70 degrees
> Celsius. I know this still might be in the green zone for the AMD Athlon 64
> (single core) its using but I'd rather keep it in even greener area of its
> threshold.
>
> I've searched the net and played with some stuff but mostly didn't find
> exactly the utility I'm looking for. I can use cpufrequtils to lower CPU
> cycles, which works ok, but why would I prefer hampering my machine's
> performance instead of simply pushing the fan cycles up?... .
>
> I took a look at lm-sensors but it has a too big config file
> (/etc/sensors.conf) for my time constraints right now and I'm not sure it'll
> help.
>
> In /proc/acpi/fan/FAN I have only "state" which when "cat"ing, prints:
> "status: on".
>
>
>  Anyone knows how to hasten the CPU speed on such a system?
>
> TIA,
>
> Boaz.
>
>
>


-- 
|
| Alex Alexander
| http://linuxized.blogspot.com
| http://www.nerd.gr
\