Save the following commands to the Firewall Script on the Administration -> Commands page to fix loopback.

insmod ipt_mark
insmod xt_mark

iptables -t mangle -A PREROUTING -i ! `get_wanface` -d `nvram get wan_ipaddr` -j MARK --set-mark 0xd001
iptables -t nat -A POSTROUTING -m mark --mark 0xd001 -j MASQUERADE

once you have added these commands to the ‘Firewall’ command list, you will need to perform a reboot and that should make the fix active.

a huge thanks to phuzi0n for providing this workaround! from the number of people who have commented, i can definitely see that i am not alone in wanting loopback to work properly but at least this workaround does the job.

while i already owned four 1 TB hard drives, two of them are western digital green drives which means they have variable RPMs; not exactly the best choice for a raid array. my main goals are speed, redundancy, and increasing my total storage space. with that being said, it seemed to me that the best choice was to buy more drives identical to my two 7200 rpm drives (hitachi deskstar 7K1000B HDT721010SLA360 to be exact). i was able to find some for a decent price ($50) from an online retailer so i picked up three. when i received my other drive back from RMA to hitachi, that would give me a total of five 1 TB drives to make a nice raid 5 array.

next, i just had to choose what file system i was going to include. different file systems provide different features and abilities. some are able to perform online resizing, some are optimized for read performance, others write, etc. overall, i was looking for something fairly reliable, as close to ‘ext3 reliable’ as possible while getting that extra boost of speed but i also wanting to pick something mainstream that i knew would be supported for years to come. to begin with, i started looking at ext4 and xfs. after doing some quick research, i realized that my server, which is running debian lenny (stable) does not include a stable ext4 driver. it only comes with ext4dev. while i could upgrade my kernel and use packages from backports, i really didn’t want to have to make that many changes to my stable, working server. so after speaking with a few colleagues and people online, i decided that xfs would be my best bet for my specific situation.

while digging further into xfs, i stumbled upon a useful mythtv wiki article regarding xfs (thanks to MRD for the link). this provided some very useful information as to how to create an optimized xfs file system for your devices. now that i had all of my information together, it was time to do a little bit of testing in a virtual environment just to familiarize myself with the process. when doing what i consider major operations, i like to test things out first. i just created a basic debian VM with multiple virtual disks, practiced creating, assembling, expanding, etc to see how well it worked.

earlier this week, my three new drives arrived. since raid 5 can run in a degraded 5 drive array with only four drives, i figured that i would go ahead and get started while still waiting for my last drive from RMA. i arranged my drives so that my raid 5 array would consist of the drives on /dev/sda to /dev/sde. many of the commands that i am using for this were found on Useful Mdadm Commands @ excaliburtech.net which is a great resource for example usage of mdadm commands. first, i created a single partition layout using fdisk on /dev/sda, set the type to ‘fd’ which is ‘Linux raid autodetect’. the layout should look like this:

root@athena:~# fdisk -l /dev/sda

Disk /dev/sda: 1000.2 GB, 1000204886016 bytes
255 heads, 63 sectors/track, 121601 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x68714c34

   Device Boot      Start         End      Blocks   Id  System
/dev/sda1               1      121601   976760001   fd  Linux raid autodetect

i could use fdisk to create the same partition layout on the other three drives but there is an easy way to copy the layout. ‘/dev/sda’ is the source disk, and the last parameter is the destination disk. this just copies the partition table layout to ensure they are identical:

sfdisk -d /dev/sda | sfdisk /dev/sdb
sfdisk -d /dev/sda | sfdisk /dev/sdc
sfdisk -d /dev/sda | sfdisk /dev/sdd

next, i opted to clean the superblocks to ensure that any existing superblock data (where mdadm stores array information for each drive) would be cleared.

mdadm --zero-superblock /dev/sd[abcd]
mdadm --zero-superblock /dev/sd[abcd]1

now we are ready to create the raid array. i already have a md0 and md1 so my next available is md2. this will create a raid 5 array using 5 devices but with one drive missing.

mdadm --create /dev/md2 --level=5 --raid-devices=5 /dev/sd[abcd]1 missing

be default, arrays are shows as read only. you can now mark the array as read-write:

mdadm --readwrite /dev/md2

you can now see the degraded array using ‘cat /proc/mdstat’:

root@athena:~# cat /proc/mdstat
Personalities : [raid1] [raid6] [raid5] [raid4]
md2 : active raid5 sda1[0] sdd1[3] sdc1[2] sdb1[1]
      3907039744 blocks level 5, 64k chunk, algorithm 2 [5/4] [UUUU_]

next, i am ready to prepare the new array with new xfs file system. using the myth wiki link above, i calculated the optimal values for my array and inserted them into the mkfs.xfs command:

mkfs.xfs -b size=4096 -d sunit=128,swidth=512 /dev/md2

now that the xfs file system has been created, i can do a test mount:

mkdir /media/md2
mount -o sunit=128,swidth=512 /dev/md2 /media/md2

my test mount was successful so i wanted to automatically mount my new array at boot so i just needed to add the following to /etc/fstab:

/dev/md2        /media/md2      xfs             sunit=128,swidth=512,noatime,nodiratime 0       0

now i also want to add the new raid 5 array’s information to /etc/mdadm/mdadm.conf

mdadm --examine --scan
ARRAY /dev/md2 level=raid5 num-devices=5 UUID=512145fc:e9a0cb9e:affa4567:83059fad

so i add it to the last line and the contents of the entire file look like:

# mdadm.conf
#
# Please refer to mdadm.conf(5) for information about this file.
#

# by default, scan all partitions (/proc/partitions) for MD superblocks.
# alternatively, specify devices to scan, using wildcards if desired.
DEVICE partitions

# auto-create devices with Debian standard permissions
CREATE owner=root group=disk mode=0660 auto=yes

# automatically tag new arrays as belonging to the local system
HOMEHOST 

# instruct the monitoring daemon where to send mail alerts
MAILADDR root

# definitions of existing MD arrays
ARRAY /dev/md0 level=raid1 num-devices=2 metadata=0.90 UUID=ad56dba9:a587a6ce:63b74a7d:37f4f41d
ARRAY /dev/md1 level=raid1 num-devices=2 metadata=0.90 UUID=9ee58430:486e6421:6e7df9ee:51ed0ff2
ARRAY /dev/md2 level=raid5 num-devices=5 metadata=0.90 UUID=512145fc:e9a0cb9e:affa4567:83059fad

i then went ahead and copied my data to the new array and two days later, my new drive arrived. after installing it in my system, i needed to create the partition information from an existing drive, clean the superblock, and add the drive to the array using the following commands:

copy partition information from an existing disk:

sfdisk -d /dev/sda | sfdisk /dev/sde

clean the superblock:

mdadm --zero-superblock /dev/sd[e]
mdadm --zero-superblock /dev/sd[e]1

add the new drive to the existing raid 5 array:

mdadm --add /dev/md2 /dev/sd[e]1

check drive synchronization status:

root@athena:~# cat /proc/mdstat
Personalities : [raid1] [raid6] [raid5] [raid4]
md2 : active raid5 sde1[5] sda1[0] sdd1[3] sdc1[2] sdb1[1]
      3907039744 blocks level 5, 64k chunk, algorithm 2 [5/4] [UUUU_]
      [============>........]  recovery = 60.4% (590546092/976759936) finish=100.7min speed=63877K/sec

now my raid 5 array is completely synchronized and is fully redundant:

root@athena:~# cat /proc/mdstat
Personalities : [raid1] [raid6] [raid5] [raid4]
md2 : active raid5 sde1[4] sda1[0] sdd1[3] sdc1[2] sdb1[1]
      3907039744 blocks level 5, 64k chunk, algorithm 2 [5/5] [UUUUU]

1. if you haven’t yet, i would suggest removing the official adobe flash package which is likely to be out of date:

   sudo aptitude remove --purge flashplugin-nonfree

2. download the latest 64 bit plugin from here, make sure to select ‘Download plug-in for 64-bit Linux’ (backup mirror). i would suggest saving the file to /tmp.
3. open a command line window and navigate to /tmp:

   cd /tmp

4. untar the archive:

   tar zxvf flashplayer_square_p1_64bit_linux_091510.tar.gz

5. if you haven’t followed my previous article, you will need to create the destination directory first:

   sudo mkdir /usr/lib/adobe-flash

   next, copy the newly extracted ‘libflashplayer.so’ to the plugin installation location:

   sudo cp libflashplayer.so /usr/lib/adobe-flash/libflashplayer_testing.so

   it might be a good idea to verify that you have moved the file properly:

   mbentley@athena:/tmp$ ls -l /usr/lib/adobe-flash/
   total 19736
   -rwxr-xr-x 1 root root  9570824 2010-09-17 09:18 libflashplayer.so
   -rw-r--r-- 1 root root 10601968 2010-09-17 10:01 libflashplayer_testing.so

6. next we are going to need to add a new entry to update-alternatives for the testing version:

   sudo update-alternatives --install /usr/lib/mozilla/plugins/flash-mozilla.so \
   flash-mozilla.so /usr/lib/adobe-flash/libflashplayer_testing.so 15

   ‘–install’ indicates that we are adding an entry, ‘/usr/lib/mozilla/plugins/flash-mozilla.so’ indicates the symlink we want to add an option to, ‘flash-mozilla.so’ indicates the update-alternatives name, ‘/usr/lib/adobe-flash/libflashplayer_testing.so’ indicates the file which you are adding, ’15′ indicates the priority (higher the number, the higher the priority).

7. now we need to set the new testing version of flash player to be current active version:

   sudo update-alternatives --config flash-mozilla.so

   make sure to choose the option for the newly created ‘/usr/lib/adobe-flash/libflashplayer_testing.so’:

   There are 2 alternatives which provide `flash-mozilla.so'.
   
     Selection    Alternative
   -----------------------------------------------
   *+        1    /usr/lib/adobe-flash/libflashplayer.so
             2    /usr/lib/adobe-flash/libflashplayer_testing.so
   
   Press enter to keep the default[*], or type selection number: 2
   Using '/usr/lib/adobe-flash/libflashplayer_testing.so' to provide 'flash-mozilla.so'.

8. we should now clean up our temporary files as we do not need them anymore:

   rm /tmp/flashplayer_square_p1_64bit_linux_091510.tar.gz /tmp/libflashplayer.so

9. after you have done that, you will need to restart your web browser and then you should have version 10.2.161.22 installed. you can verify this by going to adobe’s flash test page and looking in the ‘version information’ section.

in the future, you will simply need to execute steps 2-5 to upgrade to newer versions. if you want to install a new ‘stable’ plugin, you will just need to change step 5 to copy to ‘/usr/lib/adobe-flash/libflashplayer.so’. if you do so, you will also want to verify that update-alternatives is pointing to the correct version that you would like to use.

if you have any questions or problems, feel free to post in the comments and i will try to reply to them as soon as possible.

• first we want to remove the official ‘flashplugin-nonfree’ package:

  sudo aptitude remove --purge flashplugin-nonfree

• download libflashplayer.so from here.
• extract the file to a temporary location using:

  tar zxvf libflashplayer-10.0.45.2.linux-x86_64.so_.tar.gz

  you should now see ‘libflashplayer.so’

• from the same temporary location, execute the following commands to make a common location for the file to be stored and then to copy the file there:

  sudo mkdir /usr/lib/adobe-flash
  sudo cp libflashplayer.so /usr/lib/adobe-flash/

• next, you will need to create an update-alternatives entry using the following command (copy and paste both lines and insert it on the command line as a single command:

  sudo update-alternatives --install /usr/lib/mozilla/plugins/flash-mozilla.so \
  flash-mozilla.so /usr/lib/adobe-flash/libflashplayer.so 20

  ‘–install’ indicates that we are adding an entry, ‘/usr/lib/mozilla/plugins/flash-mozilla.so’ indicates the symlink we want to add an option to, ‘flash-mozilla.so’ indicates the update-alternatives name, ‘/usr/lib/adobe-flash/libflashplayer.so’ indicates the file which you are adding, ’20′ indicates the priority.

• now we just need to set the new flash player to be the default:

  sudo update-alternatives --config flash-mozilla.so

  make sure to choose the option for the newly created ‘/usr/lib/adobe-flash/libflashplayer.so’:

  There are 2 alternatives which provide `flash-mozilla.so'.
  
    Selection    Alternative
  -----------------------------------------------
            1    /usr/lib/gnash/libgnashplugin.so
  *+        2    /usr/lib/adobe-flash/libflashplayer.so
  
  Press enter to keep the default[*], or type selection number: 2
  Using '/usr/lib/adobe-flash/libflashplayer.so' to provide 'flash-mozilla.so'.

• now open up iceweasel (firefox) and navigate to ‘http://kb2.adobe.com/cps/155/tn_15507.html‘ and it should show the version of flash as ‘LNX 10,0,45,2′.

and that should be it! if you want to navigate to any flash based site or adobe’s flash test page you should see the plugin working. this has currently been tested on debian lenny (stable) and debian squeeze (testing).

before you start – this will only work if your photos actually exist on your device but do not show up in in the camera roll. you can verify that photos exist on your device by looking graph at the bottom of iTunes and by checking your device for the actual picture files. if you post comments about having problems but don’t mention that you haven’t verified that your photos are still available on your device, i’m sorry but i won’t be able to help you.

so i decided to upgrade to the new iOS 4 on my iphone 3gs to see what it is all about. everything went smoothly. i installed itunes 9.2, downloaded the new firmware, asked my phone to do a restore using the new firmware and then had it restore my latest backup of info. everything seemed perfect after syncing with itunes but then i went into my camera roll and noticed i had no pictures! in itunes, it showed over 1 GB worth of pictures so why were they not showing up? i searched online for a little while i didn’t really come up with much at first. many different solutions: do a backup, sync, reboot, etc but nothing worked. then i came across this discussion on the apple forums. i have slightly modified the steps. if you use windows or osx, you can download iphone explorer and follow these steps:

these steps are out of date.  see below for updated steps.
1.) connect your iphone using the usb cable
2.) launch iphone explorer
3.) backup and then delete the following files (backup instructions):
     /DCIM/.MISC/Info.plist
     /PhotoData/Photos.sqlite
     /PhotoData/PhotosAux.sqlite
4.) reboot your iphone

a recent update to iphone explorer may have changed the folder paths. if you don’t see the folders mentioned, try the following:

1.) connect your iphone using the usb cable
2.) launch iphone explorer
3.) backup and then delete the following files (backup instructions):
     /Media/DCIM/.MISC/Info.plist
     /Media/PhotoData/MISC/DCIM_APPLE.plist
     /Media/PhotoData/Photos.sqlite (or /PhotoData/Photos.sqlite)
     /Media/PhotoData/PhotosAux.sqlite (or /PhotoData/PhotosAux.sqlite)
4.) reboot your iphone

note: if you don’t have the file PhotosAux.sqlite, that’s not a problem. It is a database of the location of where photos were taken which is extracted from the EXIF data of the pictures. i would guess that it is uses by the Places feature in the camera roll.

if your phone is jailbroken and you are in ‘root’ view in iphone explorer, these instructions & file locations should work:

1.) connect your iphone using the usb cable
2.) launch iphone explorer
3.) backup and then delete the following files:
     /private/var/mobile/Media/PhotoData/.MISC/Info.plist
     /private/var/mobile/Media/PhotoData/MISC/DCIM_APPLE.plist
     /private/var/mobile/Media/PhotoData/Photos.sqlite
     /private/var/mobile/Media/PhotoData/PhotosAux.sqlite
4.) reboot your iphone

if you have linux, you shouldn’t need anything as long as you can mount your iphone as a device and delete the files listed above. i used debian (squeeze) to access the files as it automatically mounts my iphone when i plug it in. i just enabled ‘view hidden files’ and i could access the files to delete.

once you have done that, just open your camera roll and it should re-index all of your photos!

*edit: as Phil has pointed out in the comments, if the file ‘/DCIM/.MISC/Info.plist’ does not exist or if you can’t find the folder ‘/DCIM/.MISC’, just skip past that file and follow the rest of the directions and it should still restore your photos in the camera roll.

*edit: has anyone had this happen with the latest update to iOS 5? i am curious whether or not it is happening again with the latest iOS. if it is, please let me know in the comments. thanks!

yesterday i received my soekris engineering net4801-60 in the mail. the unit seems to be in good condition and even came with a compactflash card and vpn1411 card which i won’t end up using. i will most likely end up selling the vpn card as i don’t use my smoothwall to create a vpn and the smoothwall os does not have support built into the kernel for a vpn card such as this anyway.

as you can see, it really is just a basic computer. today i received the last parts that i will be needing for this project. they are the hard drive mounting kit and the power supply. i purchased these directly from soekris engineering. in part 1, i prepared the smoothwall install on a 2.5″ 60GB laptop hard drive. the next step is to install the hard drive on the mounting kit inside the unit. the hard drive connects using four mounting screws on the bottom of the drive. the mounting kit is connected to by unscrewing the four screws which secure the net4801 to the case and placing the mounting bracket on top of the board’s mounting holes and reinserting the screws into the mounts.

now that the hard drive is connected and the unit is closed back up, it is time to test the unit. in order to do that, i am going to fire up minicom on my server which is connected via null modem serial cable to the net4801 and listening on com1. if the unit does not show any activity in the minicom windows, you might want to check out the manual from the soekris download page to make sure that you have configured minicom to use the default settings (19200 baud, 8 databits, no parity, 1 stop bit, no flow control). here is the boot screen that you should see upon powering up the unit:

comBIOS ver. 1.31  20070408  Copyright (C) 2000-2007 Soekris Engineering.

net4801

0256 Mbyte Memory                        CPU Geode SC1100 267 Mhz 

Pri Mas  IC25N060ATMR04-0                LBA Xlt 1024-255-63  58605 Mbyte

Slot   Vend Dev  ClassRev Cmd  Stat CL LT HT  Base1    Base2   Int
-------------------------------------------------------------------
0:00:0 1078 0001 06000000 0107 0280 00 00 00 00000000 00000000
0:06:0 100B 0020 02000000 0107 0290 00 3F 00 0000E101 A0000000 10
0:07:0 100B 0020 02000000 0107 0290 00 3F 00 0000E201 A0001000 10
0:08:0 100B 0020 02000000 0107 0290 00 3F 00 0000E301 A0002000 10
0:18:2 100B 0502 01018001 0005 0280 00 00 00 00000000 00000000
0:19:0 0E11 A0F8 0C031008 0117 0280 08 38 00 A0003000 00000000 11

 1 Seconds to automatic boot.   Press Ctrl-P for entering Monitor.

on minicom, the unit shows a ram count and then after a few seconds proceeds to boot. due to the configuration of the kernel smoothwall uses, you can’t see the boot messages (and we didn’t set lilo up to show them even if the kernel supported it). once the unit is ready, it will display a login prompt:

smoothwall
Connected on ttyS0 at 19200 bps
smoothwall login:

from there you need to login using ‘root’ and the password you set up in the initial configuration. once you login, you will need to reconfigure your smoothwall’s network settings. you do this by running the ‘setup’ command. on the setup screen, you will want to choose ‘networking’, then ‘network configuration type’. on the next screen, choose “green + orange + red”. you will then want to choose ‘drivers and card assignments’. then choose ‘ok’ to change settings then choose ‘probe’. as it detects the network cards, set them as green, then orange, and red (the order is important). green (internal) is eth0, orange (dmz) is eth1, and red (external) is eth2. make sure the address settings are still correct for your network and your external network is set properly for your ISP.

once that is done, i connect network cables to the net4801 and then reboot the unit. once it reboots, you might want to check to make sure all of your interfaces have ip addresses and have been assigned correctly. if they have not, you will want to re-run setup and check for errors in your configuration. from there, i was able to connect to the web interface and configure my smoothwall. now i am up and running behind my net4801. from looking at my mrtg graphs, i can get an idea of how much less power the net4801 uses. my estimated runtime on battery went from ~30 to ~38 minutes.

and that’s pretty much about it for the setup. if you have any questions, feel free to post them in the comments section.

• soekris engineering net4801-60
• soekris engineering 2.5″ hdd mounting kit
• soekris engineering power supply
• 2.5″ ide hard drive
• 10 foot serial null modem cable

since i already have the hard drive, i can do some of the prep work by installing smoothwall. i am using an old laptop to install the os to the hard drive. luckily linux is much less picky than windows about changing hardware after installing. i downloaded and burnt a copy of the latest smoothwall 3.0 sp1. installing smoothwall is about as straight forward as things come. just follow the steps on the screen until you get to the “Network configuration menu”. since we currently do not have the correct network cards in the computer, this part is going to be overwritten later.

just leave the current config as “GREEN (RED is modem/ISDN)”. choose “Drivers and card assignments”. allocate the nic to the green interface. then go to “Address settings” and give the green interface a valid IP address on your subnet. once that is completed, choose done. on the “Section menu” screen, just choose finished as we do not want to set up a modem that we actually won’t be using. then just complete the setup by entering your passwords and reboot.

when your smoothwall boots, go ahead a login using “root” and the password that you provided. we need to make a few configuration changes since the net4801 only uses serial console for video output. first, we need to edit ‘/etc/inittab’. once finished, your inittab should look like this:

id:3:initdefault:

l0:0:wait:/etc/rc.d/rc.halt halt
l6:6:wait:/etc/rc.d/rc.halt reboot

si::sysinit:/etc/rc.d/rc.sysinit

# Trap CTRL-ALT-DELETE
ca::ctrlaltdel:/sbin/shutdown -h now

# Run gettys in standard runlevels
1:2345:respawn:/sbin/agetty 38400 tty1
#2:2345:respawn:/sbin/agetty 38400 tty2
#3:2345:respawn:/sbin/agetty 38400 tty3
#4:2345:respawn:/sbin/agetty 38400 tty4
#5:2345:respawn:/sbin/agetty 38400 tty5
#6:2345:respawn:/sbin/agetty 38400 tty6
s0:2345:respawn:/sbin/agetty -L -f /etc/issueserial 19200 ttyS0 vt100
~:S:wait:/bin/bash

the line in italics ‘s0:2345:respawn:/sbin/agetty -L -f /etc/issueserial 19200 ttyS0 vt100‘ is what we are adding. this enables the serial console at the baud rate of 19200 (the default for the net4801). the lines in bold are ones that we should comment out. there is no point in running multiple virtual terminals if we aren’t going to use them. they will just use resources since this box is headless but we are leaving one just in case it is needed in a recovery scenario.

next we need to create an ‘/etc/issueserial’ file:

smoothwall
Connected on \l at \b bps

next, we will want to create an ‘/etc/securetty’ file to allow root to login from the serial console:

ttyS0
tty1
tty2
tty3

in my next article, i am going to continue by talking about how to reconfigure the smoothwall setup to accept the new hardware of the net4801 but i will do so when i actually have the hardware on hand later this week.

continued in part 2

at home, i have five virtual machines that i rely on heavily for my day-to-day operations. i have two windows server 2008 domain controllers, two windows server 2003 servers running exchange 2003 and one windows server 2008 computer i use for remote desktop and management of services. some of these services can provide a fair load on my server and i was starting to see the impact, especially from my win2k8 vms. it seems that there are a number of optimizations that are not enabled by default in vmware server. i searched far and wide through vmware kb docs as well as blogs and web forums and came up with the following optimizations:

force virtual machines to run in shared memory (in ram)
inside ‘/etc/vmware/config’ lies the default config for vmware. add the following lines at the end:

MemTrimRate = "0"
mainMem.useNamedFile = "FALSE"
sched.mem.pshare.enable = "FALSE"
tmpDirectory="/dev/shm"

this will ensure that virtual machines actually store their virtual memory on /dev/shm. /dev/shm is a location that most linux distributions use for shared memory that is located directly on ram (think of a ramdisk). you will want to make sure that your host makes uses of /dev/shm as seen below:

root@athena:~# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/md0               67G  7.3G   56G  12% /
tmpfs                 4.0G     0  4.0G   0% /lib/init/rw
udev                   10M  972K  9.1M  10% /dev
tmpfs                 8.0G  5.1G  3.0G  64% /dev/shm
/dev/md2              917G  665G  207G  77% /media/md2
/dev/md3              917G  556G  315G  64% /media/md3

the text in italics above is what you are looking for. by default, most distros will not allocate that much space for shared memory. however, you can specify the size in the file ‘/etc/default/tmpfs’:

# SHM_SIZE sets the maximum size (in bytes) that the /dev/shm tmpfs can use.
# If this is not set then the size defaults to the value of TMPFS_SIZE
# if that is set; otherwise to the kernel's default.
#
# The size will be rounded down to a multiple of the page size, 4096 bytes.
SHM_SIZE=8G

by default the variable seen above will be blank (ie – ‘SHM_SIZE=’). i set mine to 8GB because i have 8GB of ram in my server. the great thing about using shared memory is that if you happen to actually fill up your system’s ram, it will start using swap space instead of just running out of room.

next, i added a few lines to the file ‘/etc/sysctl.conf’:

vm.swappiness = 0
vm.overcommit_memory = 1
vm.dirty_background_ratio = 5
vm.dirty_ratio = 10
vm.dirty_expire_centisecs = 1000

it is my understanding that these are meant to help prevent vmware from using the swap file unnecessarily.

last but not least, i needed to do a little bit of optimization of my file system. my server uses software raid 1 with an ext3 file system so write performance isn’t amazing but to speed things up, i changed a few of the mount options in my ‘/etc/fstab’:

 /etc/fstab: static file system information.
#
# <file system> <mount point>   <type>          <options>                       <dump>  <pass>
proc            /proc           proc            defaults                        0       0
/dev/md0        /               ext3            errors=remount-ro,relatime      0       1
/dev/md1        none            swap            sw                              0       0
/dev/scd0       /media/cdrom0   udf,iso9660     user,noauto                     0       0
/dev/md2        /media/md2      ext3            defaults,noatime,nodiratime     0       2
/dev/md3        /media/md3      ext3            defaults,noatime,nodiratime     0       2

i added the ‘noatime,nodiratime’ options so that write times are not constantly being recorded causing excessive delays.

these settings seem to make my virtual machines run smoother and boot much faster.

this is basically a 586 class pc that runs headless and uses a serial console for access. i thought about buying one directly from the manufacturer but then i saw the prices and they seemed more expensive (new units start at $213.00) than i wanted for such a small project. luckily i was able to find a used unit that i bought on ebay for $96. not only did this cost less than half of the price of a new unit, but it was a *-60 model instead of the *-40 they are selling now (266 mhz proc vs 233 mhz, 256mb of ram vs 128mb).

along with the unit itself, i purchased a 2.5″ hdd mounting kit since i didn’t want to go with a compactflash solution and i already have an extra 2.5″ ide hdd. hopefully i should receive everything sometime next week. i haven’t used any devices that use a serial console since i worked with cisco appliances in college but i ran a trial run of how i would install everything in vmware and it seemed to go easy enough. i’ll do a more detailed breakdown of how to install an os on a unit that is headless and has no optical drive for installation once i have everything on hand.