I Do Linux

Monit for Easy Server Process Monitoring

2009-11-05T12:53:00.015-05:00

Monit is a free open source utility for managing and monitoring, processes, files, directories and filesystems on a UNIX system. Monit conducts automatic maintenance and repair and can execute meaningful causal actions in error situations.
http://mmonit.com/monit/

Everyone has those daemons that die under pressure, which results in a late night phone call, logging into the server and restarting the process. Monit is an easy to deploy process watchdog that will restart those annoying processes. It also has a whole slew of other monitoring and alert functions.

Here, I am yum install'ing from the Dag Wieers repo.

# yum install monit
# chkconfig monit on
# service monit start

Take a look at /etc/monit.conf for some config examples, then
create your own included config file, per service you want to monitor, in the /etc/monit.d/ directory. My problem daemon is a flexlm license manager.

/etc/monit.d/flexnet:

  check process flexnet with pidfile /var/tmp/flexnet.pid
    start program = "/etc/init.d/flexnet start"
    stop program  = "/etc/init.d/flexnet stop"
    if 3 restarts within 5 cycles then timeout

My quick hack to create a pid file for flexlm with the flexnet init script /etc/init.d/flexnet:

#!/bin/sh

case "$1" in
  start)
        if [ -f /etc/lmboot_TMW ]; then
            # pid cleanup
            rm -f /var/tmp/.flexlm/lmgrd.* /var/tmp/flexnet.pid
            # start
            /etc/lmboot_TMW -u flexlm && echo 'MATLAB_lmgrd'
            # pid hack
            cat /var/tmp/.flexlm/lmgrd.* | grep PID | tail -n1 | cut -d"=" -f2 > /var/tmp/flexnet.pid
        fi
        ;;
  stop)
        if [ -f /etc/lmdown_TMW ]; then
            /etc/lmdown_TMW  > /dev/null 2>&1
            # pid cleanup
            rm -f /var/tmp/.flexlm/lmgrd.* /var/tmp/flexnet.pid
        fi
        ;;
  *)
        echo "Usage: $0 {start|stop}"
        exit 1
        ;;
esac

exit 0

reload the service after adding you files:

# service monit reload

This is just to get me started. I think I'll monitor some more services, including node uptime, and try the web interface.

GreenIT on my mind

2009-11-03T14:20:00.006-05:00

EDUCAUSE Greening Technology in U.K. Higher Education
EDUCAUSE Green IT Best Practices at the University of Michigan
Lawrence Berkeley National Laboratory’s Data Center Energy Management Technical Best Practices

Be sure to catch Joyce Dickerson's EDUCAUSE 2009 Online session.

Mount Disk Image Partition with Loop Device

2009-10-30T14:06:00.004-04:00

I didn't have enough memory to start up an old Xen VM for a file restore, so I mounted the disk image through a loop device instead.

using just losetup and mount with simple partitions:

# losetup /dev/loop7 /var/lib/xen/images/vm01.img
# fdisk -lu /dev/loop7

Disk /dev/loop7: 107.3 GB, 107374182400 bytes
255 heads, 63 sectors/track, 13054 cylinders, total 209715200 sectors
Units = sectors of 1 * 512 = 512 bytes

      Device Boot      Start         End      Blocks   Id  System
/dev/loop7p1   *          63      208844      104391   83  Linux
/dev/loop7p2          208845   205631999   102711577+  83  Linux
/dev/loop7p3       205632000   209712509     2040255   82  Linux swap / Solaris

# mount -o ro,offset=$((208845 * 512)) /dev/loop7 /mnt
# losetup -d /dev/loop7
# umount /mnt

using kpartx for lvm and extended partitions:


# kpartx -v -a /var/lib/xen/images/vm01.img
add map loop1p1 : 0 208782 linear /dev/loop1 63
add map loop1p2 : 0 205423155 linear /dev/loop1 208845
add map loop1p3 : 0 4080510 linear /dev/loop1 205632000
# mount -o ro /dev/mapper/loop1p2 /mnt

Linus Endorses Windows 7

2009-10-23T10:13:00.004-04:00

No. Not really.

Create DOS Boot Disk for CD or GRUB

2009-10-22T09:37:00.030-04:00

Sometimes hardware firmware updates are not available as simple programs you can execute while the OS is running. You have to boot into DOS to flash a motherboard BIOS chip or RAID controller. Here we will take a DOS utility and BIOS update image, copy them onto a DOS boot disk image, and then boot into it, by burning a CD-R or via a GRUB menu option. This is handy because most new hardware comes without a floppy drive.

Download the DOS floppy image and copy the utility in:

# wget http://www.fdos.org/bootdisks/autogen/FDOEM.144.imz
# unzip FDOEM.144.imz
# mount -o loop -t vfat FDOEM.144 /mnt
# cp AWDFLASH.EXE SN78U10Y.BIN /mnt/
# umount /mnt

Now, IF you still have a floppy drive, you can write this to a 1.44 floppy diskette:

# dd if=FDOEM.144 of=/dev/fd0 count=1 bs=1440k

To create a bootable CD-ROM from a floppy disk image:

# mkdir ./t
# cp FDOEM.144 ./t
# mkisofs -r -b FDOEM.144 -c boot.cat -o ./bootcd.iso ./t
# cdrecord -dao -overburn dev=/dev/sr0 speed=8 driveropts=burnfree -vvvv bootcd.iso

If Linux is already installed, I prefer to use a GRUB menu item at boot and not worry about possible external media errors. No need to burn a disc or make a floppy!

# cat FDOEM.144 | gzip > /boot/dosboot.img.gz
# cp /usr/share/syslinux/memdisk /boot/

and add the following to the bottom of your /boot/grub/grub.conf file:

title DOS update utility
   root  (hd0,0)
   kernel /memdisk
   initrd /dosboot.img.gz

CentOS 5.4 Released

2009-10-22T09:24:00.004-04:00

The latest version of the Community Enterprise Operating System has been released. Check out the mailing list announcement, read the release notes and then go download the iso from your local mirror. Remember to seed the torrent.

IBM Storage Update x3550 / MegaRAID 8480 / EXP3000

2009-10-20T11:00:00.006-04:00

Along with a recent storage upgrade, from 12TB to 36TB, I decided to do a full firmware update on an IBM x3550 with SAS direct-attached EXP3000 disk enclosures. These notes are specific to my hardware. Please contact your IBM support specialist for advice at 800-IBM-SERV.

x3550, M/T 7978, backplane, PHY Settings and AMSU Updates ??? internal only

# mkdir ibm_fw_sasphy_090217_linux-gb_32-64
# cd ibm_fw_sasphy_090217_linux-gb_32-64
# unzip ../ibm_fw_sasphy_090217_linux-gb_32-64.zip
# cp /usr/share/syslinux/memdisk /usr/lib/syslinux/
# ./ibm_fw_sasphy_090217_linux-gb_32-64.sh -s
# reboot

megaraid linux os driver MIGR-5073142

# cp /etc/redhat-release /etc/redhat-release.centos
# echo "Red Hat Enterprise Linux Server release 5.3 (Tikanga)" > /etc/redhat-release
# mkdir ibm_dd_sraidmr_00.00.03.23_rhel5_32-64
# cd ibm_dd_sraidmr_00.00.03.23_rhel5_32-64
# tar xzvf ../ibm_dd_sraidmr_00.00.03.23_rhel5_32-64.tgz
# ./install.sh

ibm hba exp3000 enclosure firmware bootable iso update MIGR-5078698 ibm_fw_esm_1.4_anyos_anycpu.iso

hard drive firmware bootable dos iso update MIGR-62832 ibm_fw_sas_hdd_v105_anyos_noarch.iso

Many updates are also bundled into the UpdateXpress System Pack for your model, currently MIGR-5080671 v1.40 for the x3550.

lsi megaraid 8480E firmware update MIGR-5069802
and management software iso MIGR-65490 lsi_sw_megasas_1.03_anyos_noarch.iso

MegaRAID SAS 8480E support downloads are also available directly from LSI, but these are not official IBM options.

# cd /usr/local/src/
# wget http://www.lsi.com/DistributionSystem/AssetDocument/support/downloads/megaraid/miscellaneous/linux/1.01.39_Linux_Cli.zip
# wget http://www.lsi.com/DistributionSystem/AssetDocument/7.0.1-0066_SAS_1068_FW_Image_APP-1.12.230-0598.zip
# unzip 1.01.39_Linux_Cli.zip
# rpm -ihv MegaCli-1.01.39-0.i386.rpm
# unzip 7.0.1-0066_SAS_1068_FW_Image_APP-1.12.230-0598
# /opt/MegaRAID/MegaCli/MegaCli64 -adpfwflash -f SAS1068_FW_Image.rom -a0

Quick Install of Dell OpenManage

2009-10-13T09:00:00.002-04:00

Dell OpenManage provides a way to remotely see the health of your direct-attached storage and monitor the general hardware health of your entire server. It is also a handy debug tool if you have to contact support. It is easy to install and worth the effort. Here we are installing / upgrading the latest version of OpenManage, currently version 6.1.0.

Goto support.dell.com downloads, enter your service tag, select "Red Hat Enterprise Linux 5" as your OS, then expand the "+Systems Management" tree and click "Dell OpenManage Server Administrator Managed Node", then select "GNU-Zip OM_6.1.0_ManNode_A00.tar.gz". Once you finally see the "Download" button, you should be able to right-click on it and "Copy Link Location" so that you can paste it into your server ssh terminal for a direct download.

# cd /usr/local/src
# wget http://ftp.us.dell.com/sysman/OM_6.1.0_ManNode_A00.tar.gz
# mkdir OM_6.1.0_ManNode_A00
# cd OM_6.1.0_ManNode_A00
# tar xzvf ../OM_6.1.0_ManNode_A00.tar.gz
# ./setup.sh

The setup will automatically un-install old versions, install the new versions, and ask you if you would like to start the service. Say yes to all.

Now you should be able to navigate in a web browser to the ssl web server running on port 1311 of your server and log in (e.g. https://localhost:1311/ ). If you are not directly on the server, you will have to setup a port forward to access this service remotely and then connect with your web browser:

$ ssh -L 1311:localhost:1311 user@myserver.foo.org

Update: It looks like you can also install OMSA with the Dell Hardware Repo latest. I would uninstall existing versions before attempting this first.

# wget -q -O - http://linux.dell.com/repo/hardware/latest/bootstrap.cgi | bash
# yum install srvadmin-all

Red BASH Prompt on Error

2009-10-02T15:19:00.012-04:00

This tweak will give you a minimal BASH prompt that prepends the exit code of the last command, in red, only in the event of an error.

Add this to the end of your ~/.bashrc file:

RED='\e[0;31m'
ESC='\e[0m'
PROMPT_COMMAND='RETURN=$?; echo -ne "\033]0;${USER}@${HOSTNAME%%.*}:${PWD/#$HOME/~}"; echo -ne "\007";'
EXIT_CODE='$(if [[ $RETURN = 0 ]]; then echo -ne ""; else echo -ne "\[$RED\]$RETURN\[$ESC\] "; fi;)'
PS1="$EXIT_CODE\u@\h:\W\$ "

And give it a try:

[me@myserver ~]$ source ~/.bashrc
me@myserver:~$ cat /non-existent-file
cat: /non-existent-file: No such file or directory
1 me@myserver:~$ ps
  PID TTY          TIME CMD
11032 pts/0    00:00:01 bash
12943 pts/0    00:00:00 ps
me@myserver:~$ false
1 me@myserver:~$ true
me@myserver:~$

More info on customizing your prompt over at IBM: Tip: Prompt magic

HPC Cluster Install: Intro

2009-10-01T14:00:00.008-04:00

It's here. A shiny new rack of equipment from Dell.

Over the next few week, I hope to document the installation process in order to start a conversation about best practices, software options and general manageability. I hope you will stay tuned and be critical of my choices. Please post comments with all your pointers and suggestions.

Specs:

31x node cluster
dual-socket quadcore 2.53GHz processors
2 processors/node = 8 cores/node = 248 cores total
infiniband interconect
12TB of storage
42U Rack, 1U KMM/KVM, 3000VA UPS 208 Volt
2x PowerEdge R610 for head and storage nodes
2x PowerVault MD1000 for storage
29x PowerEdge R410 compute nodes
QDR switch & DDR HBA InfiniBand Fabric from Qlogic
PowerConnect 6248 gigE switch
off-site disk-to-disk backup

Avoid yum-complete-transaction

2009-10-01T13:11:00.010-04:00

Yes, the command that yum told me to run, yum-complete-transaction, completely destroyed my system by removing hundreds of packages, including yum itself. What started as a normal Wednesday night maintenance window update, turned into an all night re-install session.

yum-complete-transaction removed almost all RPM packages on my system with no warning prompts. The system happened to be a Xen hardware host, running multiple virtual hosts, so it hurt. This has got to be a bug, and it is one that has bitten many others:

For the first time in a while, I long for apt, which never left me hanging.

Red Hat 5.4 Released, CentOS 5.4 Soon

2009-09-21T14:54:00.006-04:00

Red Hat has announced the release of the latest version of Enterprise Linux. This means CentOS 5.4 is 2-4 weeks away.

One big feature many would like to see is XFS support. Currently, the XFS kernel module is available as a technology preview only in Red Hat Enterprise Linux x86_64, and there are no xfs utilities included. This means no 32-bit support and no mkfs.xfs for actually formatting. It looks like CentOS already includes XFS and related utilities in the CentOS Plus repo.

See the Red Hat Enterprise Linux 5.4 Release Notes to get primed.

You can also check out the new CentOS artwork, which ditches the flowers and goes back to cubes.

VMware vSphere Linux CLI Client

2009-08-26T11:21:00.007-04:00

There is indeed a command line client for Linux. Download the client after accepting the EULA here.

# yum install perl-Crypt-SSLeay
# tar xzvf VMware-vSphere-CLI-4.0.0-161974.i386.tar.gz
# cd vmware-vsphere-cli-distrib/
# ./vmware-install.pl

Crisis Averted. You Can Stop Panicking.

2009-08-05T09:47:00.003-04:00

CentOS was able to catch up with the MIA admin, Lance:

The CentOS Development team had a routine meeting today with Lance Davis in attendance. During the meeting a majority of issues were resolved immediately and a working agreement was reached with deadlines for remaining unresolved issues. There should be no impact to any CentOS users going forward.

The CentOS project is now in control of the CentOS.org and CentOS.info domains and owns all trademarks, materials, and artwork in the CentOS distributions.

We look forward to working with Lance to quickly complete all the agreed upon issues.

More information will follow soon.
- centos.org

Add Some Color to Your Terminal

2009-08-05T08:59:00.006-04:00

I like to kick it grey on black with the terminal, pro style. That way, you can actually see the highlights that some programs output as white. Then I add a complementary splash of color by adding the following to my ~/.bashrc file:

eval `dircolors`
alias ls='ls --color=yes'
alias grep='grep --color=yes'
alias less='less -R'

Note that the above aliases will not break command line scripting by adding extra characters.

I also end up tweaking my vim editor colors in my ~/.vimrc file:

" color
set background=dark
" search
set ignorecase
set smartcase
set incsearch

Thanks, Chris!

IBM XIV

2009-07-30T11:55:00.005-04:00

There has been a lot of talk about the IBM XIV storage offering recently. The product looks to be a strong contender in the research storage space, being composed of low-cost commodity hardware. IBM even hired 400 new sales people to get it out there.

Missing CentOS Admin

2009-07-30T00:20:00.005-04:00

For over a year now, Lance Davis has been out of touch with the rest of the CentOS team, being non-responsive to phone and email. The team has just release an open letter hoping for a response. The problem is that he has sole control of the domain name, irc channel, paypal donations and adsense accounts. There is also some question about his appropriation of the funding donated to the project.

I hope that all is well with Lance and he is not ill or dead. As bad as it sounds, I would rather hear that he is just ignoring the team and misusing the funding. Maybe he is just on vacation?

In the meantime, I believe that CentOS is in good shape and consistent with their updates. Worst-case-scenario may be updating accounts to multiple other team members, and the project marches on.

Reinstall GRUB Bootloader on md0

2009-07-17T10:32:00.008-04:00

After a power outage, an RHEL 4 server was stuck with nothing on the screen but "GRUB" at the top left. Rebooting had the same result. My bootloader was toast.

I put in the RHEL 4 install disc, and at the boot prompt selected the rescue option.

boot: linux rescue

Once the rescue image booted, scanned the system and mounted read/write, I attempted to reinstall grub:

# chroot /mnt/sysimage/
# grub-install /dev/sda
/dev/md0 does not have any corresponding BIOS drive
# grub-install --recheck /dev/sda
/dev/md0 does not have any corresponding BIOS drive
#

An error. The md0 mirror, which is composed of sda and sdb, was complicating things. I though maybe I could fudge the device map by adding an entry to /boot/grub/device.map:

(fd0)   /dev/fd0
(hd0)   /dev/sda
(hd1)   /dev/sdb
(hd2)   /dev/md0

# grub-install /dev/sda
The file /boot/grub/stage1 not read correctly.

That was no good either. Finally, I had to run grub manually.

# grub
grub> root (hd0,0)        
root (hd0,0)
 Filesystem type is ext2fs, partition type 0xfd
grub> setup (hd0)
setup (hd0)
 Checking if "/boot/grub/stage1" exists... no
 Checking if "/grub/stage1" exists... yes
 Checking if "/grub/stage2" exists... yes
 Checking if "/grub/e2fs_stage1_5" exists... yes
 Running "embed /grub/e2fs_stage1_5 (hd0)"...  16 sectors are embedded.
succeeded
 Running "install /grub/stage1 (hd0) (hd0)1+16 p (hd0,0)/grub/stage2 /grub/grub.conf"... succeeded
Done.
grub> quit
# reboot

Fixed. I then also installed the bootloader on sdb (hd1), just in case.

Update RHEL 4 with CentOS 4

2009-07-15T20:03:00.011-04:00

If you would like to update your Red Hat Enterprise 4 server with the latest CentOS 4 repositories, here are some step-by-step command lines.

# cd /usr/src/
# wget http://www.gtlib.gatech.edu/pub/centos/4.7/os/x86_64/CentOS/RPMS/up2date-4.7.1-17.el4.centos.x86_64.rpm
# wget http://www.gtlib.gatech.edu/pub/centos/4.7/os/x86_64/CentOS/RPMS/centos-release-4-7.x86_64.rpm
# cp /etc/sysconfig/rhn/up2date /etc/sysconfig/rhn/up2date.redhat
# rpm -ev up2date-gnome rhn-applet
# rpm -Uhv centos-release-4-7.x86_64.rpm up2date-4.7.1-17.el4.centos.x86_64.rpm
# cp /etc/sysconfig/rhn/up2date.rpmnew /etc/sysconfig/rhn/up2date.centos
# cp /etc/sysconfig/rhn/up2date.centos /etc/sysconfig/rhn/up2date
# vim /etc/sysconfig/rhn/sources
# rpm --import /usr/share/rhn/RPM-GPG-KEY
# up2date-nox --update
# up2date-nox -f kernel
# vim /etc/grub.conf
# reboot

Converting from RHEL to CentOS only makes sense on older, semi-retired machines.

Fedora 11 on the Dell Latitude 2100 Netbook

2009-07-14T19:01:00.032-04:00

I recently received a Dell Latitude 2100 netbook for testing. This thing isn't just for classrooms. It is a ruggedized netbook with excellent Linux support. I'm hard on my stuff. The rubberized exterior lets me throw it in my bag without a case and it really grips on uneven surfaces.

Specs:

10.43" x 7.36" x 0.89 - 1.57" dimensions
Intel Atom N270 (1.60 GHz, 512KB L2 Cache, 533MHz FSB) processor
6-cell battery
10.1" WSVGA (1024 x 576) LED display
Intel WiFi Link 5300 wireless
Intel GMA 950 graphics
16GB SSD drive
1GB DDR2 memory
bluetooth
3x USB, VGA, headphone and mic ports

It ran Ubuntu Linux out of the box, but I immediately switched to Fedora. In Ubuntu, I launched the "Create Dell Recovery Media" desktop icon, just in case I had to return to factory defaults. After that, I was ready to safely start my Fedora 11 install.

The first thing to do is download the Fedora 11 Live i686 iso and create usb install media with liveusb-creator.

Once you have your Fedora usb drive, plug it into the netbook, power on and hit F12, selecting boot from usb. I did mostly standard install options. When partitioning, I removed all existing partitions and did default partitioning, including ext4.

Once the installer finishes, reboot into the new system and do your first software update by opening a terminal (on the System Tools menu) and execute the following commands, which will completely update your system:

$ su -
# yum -y install yum-fastestmirror yum-presto
# yum -y update

I also added the Adobe, Skype & RPM Fusion software repositories. Here are some extra packages I install:

# yum install vim powertop thunderbird skype AdobeReader_enu \
flash-plugin stellarium frozen-bubble dosbox vnc vlc \
fortune-mod audacity-freeworld mysql-query-browser \
rdiff-backup mplayer mencoder xscreensaver-extras-gss \
tempest-gnome-screensaver samba-client openoffice.org-writer \
openoffice.org-calc openoffice.org-impress \
screen nmap wireshark-gnome iptraf strace sysstat \
thunderbird-enigmail

To install printers, like my office's Ricoh Aficio printer, I use the custom PPD files courtesy of Linux Printing.

Since this is a netbook with a space-limited screen, I consolidate all my toolbars and adjust font sizes for the best readability: 9 in Gnome & 15 in Firefox/Thunderbird. You can also save space by installing the Littlefox & Littlebird themes. And since you will want to be able to grab and move long preference windows above the top of display:

$ gconftool-2 --set /apps/compiz/plugins/move/allscreens/options/constrain_y --type bool 0

Everything worked after a fresh install of Fedora 11. Everything but the built-in microphone, but that is fixable.

There are just a few steps to getting the mic to work. You must use driver version 5.11 from the RealTek website. You can find the RealTek linux audio pack here. Once you get the download link, edit it from 5.12 to 5.11 to get the proper file.

# tar xjvf LinuxPkg_5.11.tar.bz2
# cd realtek-linux-audiopack-5.11
# tar xjvf alsa-driver-1.0.19-5.11.tar.bz2
# cd alsa-driver-1.0.19-5.11
# yum -y install gcc kernel-devel patch
# ./configure
# make
# make install

Create /etc/modprobe.d/alsa-base.conf and add the following:

alias char-major-116 snd
alias snd-card-0 snd-272
alias char-major-14 soundcore
alias sound-slot-0 snd-card-0
alias sound-service-0-0 snd-mixer-oss
alias sound-service-0-1 snd-seq-oss
alias sound-service-0-3 snd-pcm-oss
alias sound-service-0-8 snd-seq-oss
alias sound-service-0-12 snd-pcm-oss

Also, remove the PulseAudio plugin for ALSA, so you can adjust the real audio levels with alsamixer after reboot:

# yum remove alsa-plugins-pulseaudio
# reboot

A bug report has been filed with Red Hat Bugzilla. Hopefully, the updated driver will make it into the standard kernel soon.

If you're feeling adventurous, you can add some extra tweaks to boost battery life. After playing with Intel powertop, I added the following to the /etc/rc.local file:

### some power & disk saving tweaks
echo noop > /sys/block/sda/queue/scheduler 
echo 0 > /proc/sys/vm/swappiness
echo 50 > /proc/sys/vm/vfs_cache_pressure
echo 1500 > /proc/sys/vm/dirty_writeback_centisecs
echo 20 > /proc/sys/vm/dirty_ratio
echo 10 > /proc/sys/vm/dirty_background_ratio
echo 1 > /sys/devices/system/cpu/sched_smt_power_savings
#echo 10 > /sys/module/snd_hda_intel/parameters/power_save
echo 0 > /sys/module/snd_hda_intel/parameters/power_save
echo 5 > /proc/sys/vm/laptop_mode
#echo ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
#echo ondemand > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor
#cat /sys/devices/system/cpu/cpu0/cpufreq/ondemand/sampling_rate_max > /sys/devices/system/cpu/cpu0/cpufreq/ondemand/sampling_rate
echo min_power > /sys/class/scsi_host/host0/link_power_management_policy
ethtool -s eth0 wol d
for I in `find /sys -name autosuspend -exec echo {} \;` ; do echo "0" > "$I" ; done
#modprobe sdhci

If you have an external monitor display at your desk, plug in the VGA port and go dual-screen on the fly. I like to add these to my .bashrc file:

$ alias vgaon='xrandr --output LVDS1 --mode 1024x576 --rate 60 --output VGA1 --mode 1280x1024 --rate 60 --above LVDS1'
$ alias vgaoff='xrandr --output LVDS1 --auto --output VGA1 --off'
$ vgaon
$ vgaoff

Aside from the mic driver update, this has been an easy install on a great piece of hardware. The keyboard, display, video camera, and build quality are all superior to other netbooks I have tried. I would not hesitate to recommend the 2100 to anyone looking for a new netbook, whether you opt for Windows, Ubuntu or your own install of Fedora. I like it.

Google Video Chat Coming for Linux

2009-07-14T17:36:00.004-04:00

Check out the video chat for linux thread on Gmail Help Discussion. If you would be interested in testing a Linux version of Gmail voice and video, sign up.

A Method for Migrating Files and Permissions

2009-06-30T09:25:00.008-04:00

I recently had to retire an old Solaris system. It was the one that had a decade-worth of random projects placed on it. This leads to lots of files that can be a headache to handle without a skilled team and ticketing system to back you up. My part of the project was migrating the web content. To follow is a script that automated the copy of the web directory and mapped old UIDs to new UIDs, with non-colliding GIDs preserved. This does not include the 1000 line Apache httpd.conf file I had to audit and update.

Everything went surprisingly smooth, with only a handful of special cases to deal with, and a nice log file to catch those anomalous instances. New accounts were not created, but instead orphaned and chowned to a dummy user account.

Notice that I am using read into a while loop instead of a for loop from find. This is a good way to handle filenames with spaces and special characters.

migrate-www.sh:

#!/bin/bash

### uncomment afterwards and do not re-run
#exit 1

### sync files from old to new host
mkdir /www-old
rsync -aq old_server:/www/data /www-old/

### copy old auth files and manually edit groups
rsync -aq root@old_server:/etc/passwd /www-old/passwd.old
rsync -aq root@old_server:/etc/group /www-old/group.old
cat /www-old/group.old >> /etc/group
#vim /etc/group

### map old UIDs to new UIDs
#for I in `find /www-old/data`; do 
find /www-old/data | while read I; do
        echo -n "$I"
        OUID=`ls -ndl "$I" | cut -d ' ' -f 3`
        echo -n " OUID $OUID"
        OLOGIN=`grep "\:x\:$OUID\:" /www-old/passwd.old | cut -d ':' -f 1`
        echo -n " OLOGIN $OLOGIN"
        NUID=`grep ^$OLOGIN\: /etc/passwd | cut -d ':' -f 3`
        if grep -q ^$OLOGIN\: /etc/passwd; then 
                echo " NUID $NUID"
                chown $NUID "$I" 
        else
                # chown jack(1209) files to john(3572)
                case "$NUID" in
                        "1209" )
                        echo " ADOPT 3572"
                        chown 3572 "$I"
                        ;;
                # own jill(921) to jane(372)
                        "921" )
                        echo " ADOPT 372"
                        chown 372 "$I"
                        ;;
                *)
                        echo " ORPHAN 499"
                        chown 499 "$I"
                        ;;
                esac
        fi
done

### special case perm changes
# adopt to bob
chown -R 5432 /www-old/data/projectA
# new GID for group 432 is 1487
find /www-old/data -gid 432 -exec chown :1487 {} \;

Execute the scipt and redirect all output to a log file for later analysis:

# chmod +x migrate-www.sh
# ./migrate-www.sh > migrate-www.log 2>&1 &
# tail -F migrate-www.log

Fix the Domain of System Mail

2009-06-19T15:12:00.005-04:00

If your system sends you reports via email with the sender root@localhost.localdomain, there is a quick fix. add the following lines to /etc/mail/submit.mc substituting your own hostname:

define(`confDOMAIN_NAME',`yourhost.yourdomain.net')dnl
define(`always_add_domain')dnl

then reload the config and sendmail daemon:

# cd /etc/mail
# m4 submit.mc > submit.cf
# service sendmail reload

Thanks, Dareus.

Grid Engine Job Prioritization

2009-06-19T15:01:00.003-04:00

Figuring out how to balance the sharing of a computational cluster between groups and projects can be a bit of a headache. The difficult part is the logistics of defining real world resource sharing from a business perspective and then implementing them within your queue configs and code. Here comes Sun Blueprints to the rescue. Charu Chaubal has written an excellent and understandable paper on Scheduler Policies for Job Prioritization in the N1 Grid Engine 6 System. Thanks, Charu. This has helped me out enormously.

Fedora 11 in the Wild

2009-06-10T10:17:00.003-04:00

I got a big kick out of the Fedora 11 release announcement this morning. The announcement is done up in pith helmet, explorer club fashion, as if Fedora is a wild beast from some far-off continent. As you may know, most if not all of these new features will eventually make it into enterprise Linux. Give it a read.