≡ Menu

performance

Recently, I received a call from client about FTP server not working. Basically, someone was trying to
access networked file system mounts such as cifs and nfs via ftp. This is bad idea and it will result into problems because it is not supported by many applications (I'm not talking about cluster file system here). There are many issues with this kind of weird setup:

+ Permission problem
+ I/O Problem
+ Performance issue etc

Some FTP server do offer work around but I don't recommend exporting networked file system via another file system.

Bottom line, don't mix networked file system with each other.

Download Firefox 3 Release Candidate 1

Firefox 3 release candidate 1 (RC1) is available for download from official project web site. Firefox 3 is based upon Gecko 1.9 engine, which has more than 14,000 updates including some major re-architecting to provide improved performance, stability, rendering correctness, and code simplification and sustainability. Firefox 3 has been built on top of this new platform resulting in a more secure, easier to use, more personal product with a lot more under the hood to offer website and Firefox add-on developers.

Quick Firefox 3 RC1 Installation

Type the following command at a shell prompt:
$ cd /tmp && wget 'http://download.mozilla.org/?product=firefox-3.0rc1&os=linux&lang=en-US'
$ tar -jxvf firefox-3.0rc1.tar.bz2

Backup existing settings:
$ mkdir ~/backup
$ cp -avr ~/.mozilla/ ~/backup/

Now install new firefox at /opt:
$ sudo tar -jxvf firefox-3.0rc1.tar.bz2 -C /opt/
Start Firefox:
$ /opt/firefox/firefox &

firefox-3-screen-shot
(Fig.01: Firefox 3 in action [click to enlarge image])

Download FireFox 3 RC1 for Windows / Linux / OS X

A word about addons

Following addons are not compatible with FF3:

  • Firebug (beta version should work with FF3, see comments below)
  • Yslow
  • Google toolbar
  • Stumble toolbar

Updated for accuracy!

When you are using SMP (Symmetric MultiProcessing) you might want to override the kernel's process scheduling and bind a certain process to a specific CPU(s).

But what is CPU affinity?

CPU affinity is nothing but a scheduler property that "bonds" a process to a given set of CPUs on the SMP system. The Linux scheduler will honor the given CPU affinity and the process will not run on any other CPUs. Note that the Linux scheduler also supports natural CPU affinity:

The scheduler attempts to keep processes on the same CPU as long as practical for performance reasons. Therefore, forcing a specific CPU affinity is useful only in certain applications. For example, application such as Oracle (ERP apps) use # of cpus per instance licensed. You can bound Oracle to specific CPU to avoid license problem. This is a really useful on large server having 4 or 8 CPUS

Setting processor affinity for a certain task or process using taskset command

taskset is used to set or retrieve the CPU affinity of a running process given its PID or to launch a new COMMAND with a given CPU affinity. However taskset is not installed by default. You need to install schedutils (Linux scheduler utilities) package.

Install schedutils

Debian Linux:
# apt-get install schedutils
Red Hat Enterprise Linux:
# up2date schedutils
OR
# rpm -ivh schedutils*
Under latest version of Debian / Ubuntu Linux taskset is installed by default using util-linux package.

The CPU affinity is represented as a bitmask, with the lowest order bit corresponding to the first logical CPU and the highest order bit corresponding to the last logical CPU. For example:

  • 0x00000001 is processor #0 (1st processor)
  • 0x00000003 is processors #0 and #1
  • 0x00000004 is processors #2 (3rd processor)

To set the processor affinity of process 13545 to processor #0 (1st processor) type following command:
# taskset 0x00000001 -p 13545
If you find a bitmask hard to use, then you can specify a numerical list of processors instead of a bitmask using -c flag:
# taskset -c 1 -p 13545
# taskset -c 3,4 -p 13545

Where,

  • -p : Operate on an existing PID and not launch a new task (default is to launch a new task)

See also:

UPDATED for accuracy.

The importance of Linux partitions

Disk partitioning is the creation of separate divisions of a hard disk drive using partition editors such as fdisk. Once a disk is divided into several partitions, directories and files of different categories may be stored in different partitions.

Many new Linux sys admin (or Windows admin) create only two partitions / (root) and swap for entire hard drive. This is really a bad idea. You need to consider the following points while partitioning disk.

Purposes for Disk Partitioning

An operating system like Windows / Linux can be installed on a single, unpartitioned hard disk. However, the ability to divide a hard disk into multiple partitions offers some important advantages. If you are running Linux on server consider following facts:

  • Ease of use - Make it easier to recover a corrupted file system or operating system installation.
  • Performance - Smaller file systems are more efficient. You can tune file system as per application such as log or cache files. Dedicated swap partition can also improve the performance (this may not be true with latest Linux kernel 2.6).
  • Security - Separation of the operating system files from user files may result into a better and secure system. Restrict the growth of certain file systems is possible using various techniques.
  • Backup and Recovery - Easier backup and recovery.
  • Stability and efficiency - You can increase disk space efficiency by formatting disk with various block sizes. It depends upon usage. For example, if the data is lots of small files, it is better to use small block size.
  • Testing - Boot multiple operating systems such as Linux, Windows and FreeBSD from a single hard disk.


File systems that need their own partitions
PartitionPurpose
/usrThis is where most executable binaries, the kernel source tree and much documentation go.
/varThis is where spool directories such as those for mail and printing go. In addition, it contains the error log directory.
/tmpThis is where most temporary data files stored by apps.
/bootThis is where your kernel images and boot loader configuration go.
/homeThis is where users home directories go.

Let us assume you have 120 GB SCSI hard disk with / (root) and swap partitions only. One of user (may be internal or external or cracker ) runs something which eats up all your hard disk space (DoS attack). For example, consider following tiny script that user can run in /tmp directory:

#!/bin/sh
man bash > $(mktemp)
$0

Anyone can run above script via cron (if allowed), or even with nohup command:
$ nohup bad-script &

The result can be a total disaster as entire file system comes under Denial of Service attack. It will even bypass the disk quota restriction. One of our Jr. Linux sys admin created only two partition. Later poorly written application eats up all space in /var/log/. End result was memo for him (as he did not followed internal docs that has guidelines for partition setup for clients server). Bottom line create the partition on Linux server.

If you do not have a partition schema, than following attacks can take place:

  1. Runaway processes.
  2. Denial of Service attack against disk space (see above example script).
  3. Users can download or compile SUID programs in /tmp or even in /home.
  4. Performance tuning is not possible.
  5. Mounting /usr as read only not possible to improve security.
  6. All of this attack can be stopped by adding following option to /etc/fstab file:
  • nosuid - Do not set SUID/SGID access on this partition
  • nodev - Do not character or special devices on this partition
  • noexec - Do not set execution of any binaries on this partition
  • ro - Mount file system as readonly
  • quota - Enable disk quota

Please note that above options can be set only, if you have a separate partition. Make sure you create a partition as above with special option set on each partition:

  • /home - Set option nosuid, and nodev with diskquota option
  • /usr - Set option nodev
  • /tmp - Set option nodev, nosuid, noexec option must be enabled

For example entry in /etc/fstabe for /home should read as follows:

/dev/sda1  /home          ext3    defaults,nosuid,nodev 1 2

Here is mount command output from one of my OpenBSD production server:

/dev/wd0a on / type ffs (local)
/dev/wd1a on /home type ffs (local, nodev, nosuid, with quotas)
/dev/wd0d on /root type ffs (local)
/dev/wd0e on /usr type ffs (local, nodev)
/dev/wd0f on /tmp type ffs (local, nodev)
/dev/wd0h on /var type ffs (local, nodev, nosuid)
/dev/wd0g on /var/log type ffs (local, nodev)

How do I obtain information about partitions?

There are several ways that information about partitions can be obtained on Linux / UNIX like operating systems.

List partitions:

fdisk -l

Report file system disk space usage:

df -h
OR
df -k

Display partition mount options including mount points

mount
Sample output:

/dev/sda2 on / type ext3 (rw,relatime,errors=remount-ro)
tmpfs on /lib/init/rw type tmpfs (rw,nosuid,mode=0755)
/proc on /proc type proc (rw,noexec,nosuid,nodev)
sysfs on /sys type sysfs (rw,noexec,nosuid,nodev)
varrun on /var/run type tmpfs (rw,nosuid,mode=0755)
varlock on /var/lock type tmpfs (rw,noexec,nosuid,nodev,mode=1777)
udev on /dev type tmpfs (rw,mode=0755)
tmpfs on /dev/shm type tmpfs (rw,nosuid,nodev)
devpts on /dev/pts type devpts (rw,noexec,nosuid,gid=5,mode=620)
fusectl on /sys/fs/fuse/connections type fusectl (rw)
/dev/sda1 on /media/sda1 type fuseblk (rw,nosuid,nodev,allow_other,default_permissions,blksize=4096)
/dev/sda5 on /share type fuseblk (rw,nosuid,nodev,allow_other,default_permissions,blksize=4096)
/dev/sdb2 on /disk1p2 type ext3 (rw,relatime,errors=remount-ro)
securityfs on /sys/kernel/security type securityfs (rw)
debugfs on /sys/kernel/debug type debugfs (rw)
binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw,noexec,nosuid,nodev)
gvfs-fuse-daemon on /home/vivek/.gvfs type fuse.gvfs-fuse-daemon (rw,nosuid,nodev,user=vivek)

Display / edit file system configuration options

less /etc/fstab
or
vi /etc/fstab

Quickly remount /usr in ro mode

mount -o remount, ro /usr

Quickly mount all file system configured in /etc/fstab

mount -a

References: