Pages

Search

how to automatically reset linux server on hard disk failure? - Forums Linux

how to automatically reset linux server on hard disk failure? - Forums Linux


how to automatically reset linux server on hard disk failure?

Posted: 29 Jun 2009 01:47 AM PDT

> What you actually need is a custom modified kernel, with a disk driver 

Sounds good. But how to get such a driver, because I can not write it on
my own. :-) I'm hoping that I might not be the first one asking for such
a functionality. 

That's not the background of my problem. The server is fine and runs
well. But I want it to be prepared for the worst case, which means the
customer is running the server until both hard disks of its RAID1 are
defective. In this case the other cluster node has to migrate all
resources to itself, but right now the failed node is still alive enough
to make the cluster think that everything is ok.

Please don't let us argue about the fact, that no reasonable IT
department or service personal would let it come so far. Agreed. But
reality showed us more than once, that some people in fact wait until
the system is broken completely, before taking actions. Therefor I want
to harden the cluster for such cirstances.

Upgrading from CentOS 5.2 to 5.3 on a *live* system?

Posted: 28 Jun 2009 08:42 PM PDT

On Sun, 28 Jun 2009, Carlos Moreno wrote:
 

I actually tried this on several non-critical systems to see what would
happen after doing the update without rebooting. Everything went smoothly,
and there was no fallout. There is a kernel update, which of course won't
take effect until a reboot (but I did that too, several days later, with
no ill effects). From the release notes, you may need to upgrade glibc
first; the procedure that I actually used was:

yum clean all
yum update glibc\*
yum update yum\* rpm\* python\*
yum clean all
yum update kernel\*
yum update

Steve

RH6 Kernel 2.2.5-15 no input device after bootup

Posted: 25 Jun 2009 12:41 PM PDT

Thank you for your advice! But I tried to change mouse and keyboard,
checked the plugs. Everything the same.

I have to admit I was away a few days. I exported the display to another
machine running SuSE, so I can use the devices.

Unfortunately three weeks ago I bought a new Windoze laptop, a cheap
one, because my daughter wanted to play chess. I wiped the HD, so she
can not connect to the Internet and only installed a chess program.
There was a running Cygwin on it I used for accessing my private network
from far away. I used it in my days of C-programming... and felt I don't
need in any more...

Until now I didn't try Moe's diagnostic, but it seems most promising to
me. As soon as I have results I will report.

Any hints welcome. Thank you so much so far.

Ernst

Matt Giwer schrieb: 

Installing Google Chrome on Slackware

Posted: 25 Jun 2009 06:50 AM PDT

On Fri, 26 Jun 2009 03:34:28 +0200, Sidney Lambe wrote:

 

And a major source of email addresses for spammers.



--
Great advances in Debian Linux; post a bug report and get spam in three
days.







--
Great advances in Debian Linux; post a bug report and get spam in three
days.

Very strange thing.. I'm puzzled.

Posted: 23 Jun 2009 06:35 AM PDT

Zibri writes: 

And that's your problem. While Ghost claims to "copy the disk",
it doesn't. It tries to be "smart" and not copy things that "don't
matter" (don't matter to Windows, that is). In general Linux tools do a
better job of handling Windows installations than Windows tools do with
Linux installations.
--
John Hasler
gt.org
Dancing Horse Hill
Elmwood, WI USA

eth0 not found on Optiplex 760 after FC9 install

Posted: 22 Jun 2009 12:42 PM PDT

The Natural Philosopher wrote: 

See if there's an eth0 in /etc/udev/rules.d/70-persistent-net.rules.

udev shouldn't skip eth0 unless it thinks there already is an eth0.

New installation error

Posted: 20 Jun 2009 09:13 AM PDT

On Monday 29 June 2009 20:50, someone identifying as *erziggy* wrote
in /comp.os.linux.setup:/
 

In recent kernels, the */dev/hd?* nomenclature has been replaced with
SCSI nomenclature, and so the device special file for the CD-ROM would
now probably be */dev/sr0* or */dev/scd0* - the distinction between the
latter two is distro-specific, as far as I know, as it depends on
the /udev/ rules. This is customizable, though.
 

Newer distributions all make use of /udev,/ which uses a /tmpfs/ mounted
over */dev* by the init scripts at boot time. Previously a similar
thing existed with /devfs/ but this was proper to kernels prior to 2.6
and is no longer being maintained. /devfs/ used non-standard filenames
for the device special files and ran in kernelspace, whereas /udev/
creates the device special files via userspace tools, based upon the
information exported by the kernel to */sys,* which is now a separate
mountpoint in the root directory, as opposed to a subdirectory of
*/proc* as it used to be.

Distributions that shipped with /devfs/ typically still had the on-disk
*/dev* directory populated with device special files, but in /udev/
systems this is most often no longer the case, and thus */dev* would be
empty until /udevd/ is started by the init scripts. /udev/ can then
populate */dev* in two ways, i.e. either by total autodetection for all
hardware or by untarring an archive of device special files of which it
is generally assumed that they are needed on every system and only
creating the extra device special files as more hardware is detected
and/or hotplugged into the system. Likewise, hardware that is removed
from the system through hot-unplugging gets its device special files
removed.

In my humble opinion, /udev/ is mainly an effort to make GNU/Linux look
more like Windows, and in the process, it then also comes with all the
same kinds of Windows-like quirks. Officially, /udev/ was developed to
avoid the clutter of device special files typically found in on-disk
*/dev* filesystems and to facilitate hotplugging and hot-unplugging.
 

That might be a matter of loading the correct driver for that device. I
don't really have any experience with tapedrives, though.
 

Hmm... Taking a wild guess here, but I presume this has to do with
certain settings in the BIOS, preventing read access, write access or
perhaps both to a given region of memory. Something to do with caching
perhaps?

--
*Aragorn*
(registered GNU/Linux user #223157)

When to update RHEL kernels?

Posted: 19 Jun 2009 04:54 AM PDT

On Jun 19, 11:30am, Doug Freyburger <com> wrote: 

Besides using MySQL instead, or maybe SQLite for lighter databases? If
you're stuck with it, though, I'd link the kernel upgrades to OCFS2
RPM updates directly. There are some good yum utilities for doing
precisely this and blocking surprise updates.

changing partition size.

Posted: 17 Jun 2009 01:26 AM PDT

On Jun 19, 7:57pm, Nico Kadel-Garcia <com> wrote: 

Thanks everyone. resize2fs worked perfectly. The reason I used dd is
because there is some data on the disk that is not partition/file-
system specific. In fact I had tried using Acronis to clone the
partition but it didn't work. It didn't copy that data over. dd copies
everything sector by sector which is good for my purpose.

Triple booting a Macbook w/ 2 Linux systems.

Posted: 06 Jun 2009 12:57 PM PDT

piscesboy wrote:
 

The linux kernel understands MBR (DOS/MSW; PC-BIOS)
and GPT (EFI). Unfortunately, the two main boot loaders, lilo
and grub don't. There are, however, two ways to install grub
on a GPT volume. The dirty one uses the fact that GPT allows
'hidden' blocks between the partition table and the first user
LBA (stage 1_5 of grub is written in those blocks with dd(1));
grub uses the legacy MBR (LBA 0; should, according to GPT
specs, contain only one, fake entry, but linux will ignore extra
enries and use the real GPT partition table) to locate the /boot
file system. The cleaner way is to dl and aplly a set of patches
to grub, that make grub more or less GPT aware.

The MBR partition format is crap, if you don't mind the extra
effort with grub, go for GPT.

--
printf -v email $(echo \ 155 141 162 143 145 154 155 141 162 \
143 145 154 100 157 156 154 151 156 145 56 156 154 | tr \ \\)
# O Herr, lass Hirn vom Himmel fallen! #

problems with modem (ttyS0 and wvdial)

Posted: 06 Jun 2009 05:43 AM PDT

François Patte wrote: 
I dont.

Not even using wvdial.,

User cannot mount DVD, USB and other drives

Posted: 05 Jun 2009 11:12 PM PDT

On Sat, 6 Jun 2009, The Natural Philosopher wrote:

 
Mount the device, then change permission. Or more likely ownership. That
"sticks" ie the device will keep what you apply even after unmounting.
The mount point will stay as it was before you did this, so you'd
see /mnt/hd as being owned by root, but then when you mount the properly
set up drive to /mnt/hd, ownership becomes whoever you set it to.

Michael

installing jesred but getting remove kernel image

Posted: 05 Jun 2009 05:50 AM PDT

On 5 Giu, 16:49, Bill Mar <net> wrote: 


I installed Debian Lenny with Linux-image-2.6-686.
If run "apt-cache depends jesred", I got:

jesred
Depends: libc6
|Depends: squid
Depends: squid3
|Recommends: dhttpd
Recommends: <httpd>
apache2-mpm-itk
bozohttpd
tntnet
aolserver4
aolserver4-core
apache2-mpm-event
apache2-mpm-prefork
apache2-mpm-worker
boa
caudium
cherokee
dhttpd
ebhttpd
fnord
lighttpd
mathopd
micro-httpd
mini-httpd
nginx
thttpd
webfs
yaws

while if I run "dpkg -l | grep linux-image", I got:

ii linux-image-2.6-686 2.6.26+17+lenny1 Linux
2.6 image on PPro/Celeron/PII/PIII/P4
ii linux-image-2.6.26-2-686 2.6.26-15lenny2 Linux
2.6.26 image on PPro/Celeron/PII/PIII/P4


If I try to install jesred by aptitude I got:

Packages being deleted due to unsatisfied dependencies: id linux-
image-2.6-686
Packages to be removed: idA linux-image-2.6.26-2-686

glibcxx_3.4.9 not found - how to fix?

Posted: 03 Jun 2009 04:42 PM PDT

In article <invalid>, invalid
(Charlie Gibbs) writes:
 

I've tracked down and corrected down a couple of other glitches too,
and now my box not only seems to be fully operational, when I tried to
run fgrun (which got me into this mess in the first place) it came up.

And in the end, I've probably come up with the answer to my general
troubleshooting question too. Examine dmesg for error messages -
scan all of /var/log/messages if necessary. (If the system can't
boot far enough to get to a root login, boot from the installation
CD, mount the hard drive, then scan /var/log/messages.) Once you've
found a suspicious error message, type its exact text into google.
It might take some wading through results, but it's amazing what
you can find.

I'll remember that for next time.

--
/~\ invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ HTML will DEFINITELY be ignored. Join the ASCII ribbon campaign!

pre-allocate disk space within a folder

Posted: 29 May 2009 01:53 AM PDT

On Fri, 29 May 2009 01:53:43 -0700, com wrote:
 
Disk space problems are pretty much cured by Moore's law and just throwing
more hardware at the problem. However, here is one idea for implementing
a preallocated block. The block will be user writable at a specific
mount point under tmp. Perhaps, the compiler can be set to use that
subdirectory, etc.

This example uses a loopback allocation. The example below shows a
1G preallocation.

dd if=/dev/zero of=~/prealloc bs=512 count=2000000
losetup /dev/loop0 ~/prealloc
mke2fs /dev/loop0
mkdir /tmp/user_x
mount /dev/loop0 /tmp/user_x
chown user_x:users /tmp/user_x

Note: some of the above commands need root priviledges (su or sudo). If
working as root, add a chown to reset ownership, or use another method to
allow user access.

Caution: verify all commands work as expected and are appropriate for your
specific situation.

--
Douglas Mayne

Grub is hanging and won't reinstall after use of GParted

Posted: 28 May 2009 05:38 PM PDT

On May 29, 9:55pm, RyanMcCoskrie <com> wrote: 

Buy some Fedora disks, just to save trouble. You can also use them to
set up a local website installer for PXE installation. Seriously.
http://fedoraproject.org/wiki/Distribution/LocalVendors should also
list some local vendors.

Debian limits HD to 8.45GB

Posted: 25 May 2009 09:01 AM PDT


"Bryce" <invalid> wrote in message
news:gvepli$fbt$eternal-september.org... 


Sure thing, FWIW: I have seen a situation similar to yours before
and it did end up being a flaky HD...
so you may want to try another drive just to see how it reacts 


Why does it work

Posted: 19 May 2009 03:10 PM PDT

frankjg wrote: 


Funny stuff. I am technically inept but still with a little
help from my friends manage to use Linux for everything but the
digital rights monopoly protected stuff.

I write business letters and deal with my e-mail and
am able to see most sites, I make DVDs and CDs from isos. So
Linux is desktop ready in my humble opinion. Oh and I do it
on an old Dell with a 700 MHz Pentium III, running Mandriva
2008.1 and KDE 3.5.9.

later
bliss

debian kde 3.5 EN/DE keyboard Q

Posted: 16 May 2009 10:09 AM PDT

com wrote:



I've just seen your self-answer :)
--
"All science is either physics or stamp collecting."
Ernest Rutherford.
"All science is either physics AND stamp collecting".
Somewhere on the net

pxeboot kickstart with 1000 clients

Posted: 13 May 2009 10:17 PM PDT

On May 16, 12:18pm, Nico Kadel-Garcia <com> wrote:
 

I forgot to mention. There are two tricky bits to do in a '%post --
nochroot' script.

1: cp /etc/resolv.conf /mnt/sysimage/etc/resolv.conf

This sets up the missing network bit so that any network based
operations, such as 'yum update' or other downloading you want to do
in the '%post' step, work correctly.

2: cp /tmp/ks.cfg /mnt/sysimage/root/ks.cfg

This gets the *actual* kickstart file you used into the installed OS,
because that piece of misinterpreted output that anaconda puts out
called /root/anaconda-ks.cfg is fine fir recording special setups such
as X windows configurations or manual network setups or partition
setups, but it throws out a lot of the useful scripting steps and much
of the package information that you may have had in your actual
kickstart script.

And there's also some very useful final steps to do in their own
'%post' scripts.

* Start sshd. That way, if the kickstart gets stuck, you can still
access the machine remotely and fix things. It's worth the slight,
slight risk of SSH having a security flaw before being updated.

* Run 'yum -y update'. This takes a while and doesn't always work due
to conflicts that have to be resolved manually, but it's awfully
handy.

* Set up email so that it works, and send yourself a note that the
installation is complete.

Configuring CPU frequency scaling

Posted: 12 May 2009 01:49 PM PDT

Allen Kistler <moc> wrote:
 
power, 
affect 
2.6.28-12-generic, which is installed with Ubuntu 9.04
--
Richard Kimber
http://www.psr.keele.ac.uk/

Can I update libc without hassle?

Posted: 11 May 2009 05:04 AM PDT

Michael Mauch <de> wrote: 
 
 

OK. First time I tried chroot I got the error message "FATAL: kernel too
old", so I created another boot system with a newer kernel, otherwise
everything the same. I then did

chroot /mnt/hde8

, followed by

su acm

.. I was then able to start GHC and run a "hello world" command in it. :-)

Only trouble, a minor one, my console keyboard layout and character set
didn't load in the newer kernel. I can surely sort this out.

Thanks muchly!
 

--
Alan Mackenzie (Nuremberg, Germany).

Squirrelmail anyone?

Posted: 10 May 2009 12:25 PM PDT

On 05/13/09 04:03 AM, com wrote: 

I saw the name was conf.pl, but you figured it out....

Well, your alias /webmail seems to work, and you get no permission error
for config.php, so the only thing I can guess is that the apache user
has no right to read the config directory.

/bb

How do I find out what chipset I have?

Posted: 09 May 2009 11:04 AM PDT

On May 9, 7:13pm, John Reiser <net> wrote: 

Reboot, and read the motherboard's model number from the boot screen,
or check your receipts. (Dell has a fabulous website with their
service tags giving you details of your original hardware order, woo-
hoo!)