What is swap?

Swap space is the area on a hard disk which is part of the Virtual Memory of your machine, which is a combination of accessible physical memory (RAM) and the swap space. Swap space temporarily holds memory pages that are inactive. Swap space is used when your system decides that it needs physical memory for active processes and there is insufficient unused physical memory available. If the system happens to need more memory resources or space, inactive pages in physical memory are then moved to the swap space therefore freeing up that physical memory for other uses. Note that the access time for swap is slower therefore do not consider it to be a complete replacement for the physical memory. Swap space can be a dedicated swap partition (recommended), a swap file, or a combination of swap partitions and swap
files.

Why do I need swap?

Memory consuming programs Sometimes, a large program (like MySQL) make the entire system need extra memory. A significant number of the pages used by these large programs during its startup may only be used for initialization and then never used again. The system can swap out those pages and free the memory for other programs or even for the disk cache. In these cases, swap will be used to help the system handle any extra load.

Hibernation (suspend-to-disk) The hibernation feature (suspend-to-disk) writes out the contents of RAM to the swap partition before turning off the machine. Therefore, your swap partition should be at least as big as your RAM size. The hibernation implementation currently used in Ubuntu, swsusp, needs a swap or suspend partition. It cannot use a swap file on an active file system.

Unforeseeable Circumstances Unforeseeable events can and will happen (a program going crazy, some action needing much more space than you thought, or any other unpredictable combination of events). In these cases, swap can give you an extra delay to figure out what happened, or to finish what you are working on.

Symptoms

MySQL is being slow and when we run programs like ‘top’ we can see that our process (or server) is using a large quantity of swap.

What is this affecting the MySQL performance ?

MySQL will change the way works since it will replace the internal algorithms to be optimized for Disk I/O due the navy use of swap. This will make the system slower.

What we can do ?

First we need to figure out how much swap is the system using and how much free memory does it have.

1) The first trick we can do is to drop disk caches, this will free some more memory:

echo 3 > /proc/sys/vm/drop_caches

2) If we can move the swap into real memory (swap usage < free memory) we can run:

swapoff -a (to disable the swap usage, this will force to move data to memory)
swapon -a (will re-enable use of swap)

3) Check to see how aggressive is the system setup to use swap by using:

cat /proc/sys/vm/swappiness

I personally prefer to have this value around 10 (0 no not use swap  – 100 use swap)

to change this

edit /etc/sysctl.conf and add a new line “vm.swappiness=10“. Save the file and run

sysctl -f /etc/sysctl.conf

4) Make sure that mysql is allocating memory close to equal for each physical CPU (not core), you may check this by typing

numactl –hardware

due the NUMA architecture this may be one of the main issue why mysql is swapping in the first place.

Now, find the mysqld_safe file and open it with your editor, add the following line (after cmd=”$NOHUP_NICENESS”)

cmd=”/usr/bin/numactl –interleave all $cmd”

Now, when MySQL needs memory it will allocate it interleaved across all nodes, effectively balancing the amount of memory allocated in each node. This will leave some free memory in each node, allowing the Linux kernel to cache data on both nodes, thus allowing memory to be easily freed on either node just by freeing caches (as it’s supposed to work) rather than paging.

5) Edit my.cnf (or your mysql config file) to be a good match with the system (memory, buffers, threads) you’re running MySQL.

Hope that was helpful:

More resources:
The MySQL “swap insanity” problem and the effects of the NUMA architecture
Ubuntu SwapFAQ

Compile Ubuntu kernel

get the tools for building the kernel

apt-get -y install build-essential libncurses5-dev fakeroot kernel-package  linux-source

the kernel source code will be in /usr/src/

go to /usr/src and decompress the kernel

tar -xjf linux-source-2.6.28.tar.bz2

and make a symlink

ln -s /usr/src/linux-source-2.6.28 linux

now go the the kernel menu and select what options do you want to compile in your kernel

make menuconfig

next,  rebuild the kernel

cd /usr/src/linux
fakeroot make-kpkg --initrd --revision=subzero.1.0 kernel_image

Did you use mac ports, and after each update your ports are getting bigger and bigger ? Do you want to cleanup a few hundred of MB ? This is how you do it!

In order to see what is installed on your machine

port installed

to checkout the last version of the ports from the repository

sudo port -v selfupdate

howto update the ports

sudo port upgrade outdated

now to clean all the old version

sudo port clean --all installed

You can access a remote file system securely using sshfs and fuse which is a command to mount a remote filesystem encrypted through ssh.
This way you will be able to access remote files as if they were on you machine, just remember that if the connection between the computers is slow, the access will also be pretty slow

Package needed

sshfs
fuse-utils

Installation

Get the packages

For Debian: apt-get install fuse-utils sshfs
For Ubuntu: sudo apt-get install fuse-utils sshfs
For Fedora and Centos: yum install fuse-utils sshfs
For Mandriva: urpmi: urpmi fuse-utils sshfs


Next step is to mount the fuse module

modprobe fuse

Next create the mount point

mkdir /mnt/remote-fs
chown [your-user]:[your-group] /mnt/remote-fs/

Add yourself to the fuse group

adduser [your-user] fuse


Untill here all the command should be issued as root, now switch to your users and mount the remote filesystem.

sshfs remote-user@remote.server:/remote/directory /mnt/remote-fs/

It will now ask you accept the key if this is the first time you connect to that PC using ssh, and then the password, or only the password if this is not the first time you use ssh to connect to the remote server . That should be all.

First you need to install the python netifaces module:

On a debian system you can install the module like this:

sudo apt-get install python-netifaces

You’re done, now you can write the code:

from netifaces import interfaces, ifaddresses, AF_INET

for ifaceName in interfaces():
addresses = [i['addr'] for i in ifaddresses(ifaceName)[AF_INET]]
print ‘%s: %s’ % (ifaceName, ‘, ‘.join(addresses))

on my computer I have an output like this:

lo: 127.0.0.1
eth0: 1.1.1.1
eth1: 192.168.0.1

No Sense ?

An DB9/RJ45 adapter, free USB port, mac ports installed.

Installing

After the serial port is plugged in you USB port you would be able to see the adapter on Apple>About This Mac>More Info

Next we need some drivers for the adapter.

I found the serial mac driver on Future Technology Website under Virtual Serial Ports Section.

After I install the driver a new device is visible

Now the driver is ready and working.

Installing minicom (from ports)

sudo port install minicom

Howto setup minicom in order to connect to a serial port

sudo minicom –color=on -s

go to Serial port setup> and configure the serial port and data parity to look like this

Save the data (as default). and start the connection. Now you should be able to see the device on the other end of you serial connection, probably a Cisco.

RJ45 cable

This adapter works with the RJ45 serial port found on most Cisco routers. It also works on some Sun servers.
This is pretty much just like the crossover cable but with a different pinout:

1 – 6
2 – 5
3 – 3
4 – 8
5 – 7
6 – 4
7 – 1
8 – 2

- one side

- the other side

DB9 adapter

This is the trickiest part. In order to make your cable compatible with the largest number of serial devices possible, you need to combine a couple pins and split another one. Here is the pinout:

DB9 pin signal RJ45 pin color *
——- —— ——– —–
1 DCD 8 white
2 RxD 3 black
3 TxD 1 blue
4 DTR 5 green
5 SG 2 and 6 orange and yellow
6 DSR 8 white
7 RTS 4 red
8 CTS 7 brown
9 RI none

So our ethernet turned serial cable will be pinned:

1 TxD pair one
2 SG pair one
3 RxD pair two
4 RTS pair three
5 DTR pair three
6 SG pair two
7 CTS pair four
8 DSR/DCD pair four

* My DB9F/RJ45F modular adapters are colored blue, orange, black, red, green, yellow, brown, white (RJ45 1-8), but yours may not be. If yours are different, ignore the colors in the above pinout.

DB9 pins 2, 3, 4, 7, and 8 are easy. Just push the appropriate pin in the back of the DB9 connector until it snaps. Be careful not to get any of the pins mixed up because errors are a bit difficult to fix. I seem to remember having a skinny little tool to pop the pins back out again some years back, but I can’t find it anymore.

DB9 pin 5 needs two wires connected to it. Snip the pins off of the wires coming from RJ45 pins 2 and 6 (orange and yellow on mine), strip about 3 mm off the end of each, and crimp them together onto one of your spare pins.
wiring detail photo

RJ45 pin 8 has to connect to both 1 and 6 on the DB9 connector. Snip the pin off of the white wire, strip the end, cut about an inch of scrap CAT5 and pull out two of the white wires, strip both ends off of them, crimp a pin on each one, and splice all three loose ends together. You can break the end off of a spare pin and use it to crimp the three wires together. You can use the other two white wires from the inch of CAT5 for the second adapter.

Both of the DB9/RJ45 adapters should be wired exactly the same way, regardless of whether they will be used for DTE or DCE devices

The final product

That’s how our final custom cisco console cable should look: