Difference between revisions of "Improve pacman performance"

Revision as of 14:15, 13 October 2013

zh-CN:Improve Pacman Performance zh-TW:Improve Pacman Performance

Improving database access speeds

Pacman stores all package information in a collection of small files, one for each package. Improving database access speeds reduces the time taken in database-related tasks, e.g. searching packages and resolving package dependencies. The safest and easiest method is to run as root:

# pacman-optimize

This will attempt to put all the small files together in one (physical) location on the hard disk so that the hard disk head does not have to move so much when accessing all the packages. This method is safe, but is not foolproof. It depends on your filesystem, disk usage and empty space fragmentation. Another more aggressive option would be to first remove uninstalled packages from cache and to remove unused repositories before database optimization:

# pacman -Sc && pacman-optimize

Improving download speeds

Pacman's speed in downloading packages can be improved by using a different application to download packages instead of Pacman's built-in file downloader.

In all cases, make sure you have the latest Pacman before doing any modifications.

# pacman -Syu

Using Powerpill

Powerpill is a full wrapper for Pacman that uses parallel and segmented downloads to speed up the download process. Normally Pacman will download one package at a time, waiting for it to complete before beginning the next download. Powerpill takes a different approach: it tries to download as many packages as possible at once.

The Powerpill wiki page provides basic configuration and usage examples along with package and upstream links.

Using powerpill-light (deprecated)

pacman2aria2 provides a script named "powerpill-light", which was a stopgap created after the deprecation of the original powerpill written in Perl. Now that Powerpill has been re-released, powerpill-light is deprecated.

Using wget

This is also very handy if you need more powerful proxy settings than pacman's built-in capabilities.

To use wget, first install it with pacman -S wget and then modify /etc/pacman.conf by adding the following line to the [options] section:

XferCommand = /usr/bin/wget -c --passive-ftp -c %u

Instead of putting wget parameters in /etc/pacman.conf, you can also modify the wget configuration file directly (the system-wide file is /etc/wgetrc, per user files are $HOME/.wgetrc.

Using aria2

aria2 is a lightweight download utility with support for resumable and segmented HTTP/HTTPS and FTP downloads. aria2 allows for multiple and simultaneous HTTP/HTTPS and FTP connections to an Arch mirror, which should result in an increase in download speeds for both file and package retrieval.

Installation

Download and install aria2 and its dependencies:

# pacman -S aria2

Configuration

Edit /etc/pacman.conf by adding the following line to the [options] section:

XferCommand = /usr/bin/aria2c --allow-overwrite=true -c --file-allocation=none --log-level=error -m2 -x2 --max-file-not-found=5 -k5M --no-conf -Rtrue --summary-interval=60 -t5 -d / -o %o %u

Option Details

/usr/bin/aria2c

The full PATH to the aria2 executable.

--allow-overwrite=true

Restart download if a corresponding control file does not exist. (Default: false)

-c, --continue

Continue downloading a partially downloaded file if a corresponding control file exists.

--file-allocation=none

Do not pre-allocate file space before download begins. (Default: prealloc) ¹

--log-level=error

Set log level to output errors only. (Default: debug)

-m2, --max-tries=2

Make 2 maximum attempts to download specified file(s) per mirror. (Default: 5)

-x2, --max-connection-per-server=2

Set a maximum of 2 connections to each mirror per file. (Default: 1)

--max-file-not-found=5

Force download to fail if a single byte is not received within 5 attempts. (Default: 0)

-k5M, --min-split-size=5M

Only split the file if the size is larger than 2;5MB = 10MB. (Default: 20M)

--no-conf

Disable loading an aria2.conf file if it exists. (Default: ~/.aria2/aria2.conf)

-Rtrue, --remote-time=true

Apply timestamps of the remote file(s) and apply them to the local file(s). (Default: false)

--summary-interval=60

Output download progress summary every 60 seconds. (Default: 60) ²

-t5, --timeout=5

Set a 5 second timeout per mirror after a connection is established. (Default: 60)

-d, --dir

The directory to store the downloaded file(s) as specified by pacman.

-o, --out

The output file name(s) of the downloaded file(s).

%o

Variable which represents the local filename(s) as specified by pacman.

%u

Variable which represents the download URL as specified by pacman.

Additional notes

¹ --file-allocation=falloc

Recommended for newer file systems such as ext4 (with extents support), btrfs or xfs as it allocates large files (GB) almost instantly. Do not use falloc with legacy file systems such as ext3 as prealloc consumes approximately the same amount of time as standard allocation would while locking the aria2 process from proceeding to download.

² --summary-interval=0

Supresses download progress summary output and may improve overall performance. Logs will continue to be output according to the value specified in the log-level option.

³ XferCommand = /usr/bin/printf 'Downloading ' && echo %u | awk -F/ '{printf $NF}' && printf '...' && /usr/bin/aria2c -q --allow-overwrite=true -c --file-allocation=none --log-level=error -m2 --max-connection-per-server=2 --max-file-not-found=5 --min-split-size=5M --no-conf --remote-time=true --summary-interval=0 -t5 -d / -o %o %u && echo ' Complete!'

Using this XferCommand gives less useful, but much more readable, output.

pacget (aria2) mirror script

This script will greatly improve the download speed for broadband users. It uses the servers in /etc/pacman.d/mirrorlist as mirrors in aria2. What happens is that aria2 downloads a single package from multiple servers simultaneously which gives a huge boost in download speed.

/usr/bin/pacget

#!/bin/bash

msg() {
  echo ""
  echo -e "   \033[1;34m->\033[1;0m \033[1;1m${1}\033[1;0m" >&2
}

error() {
  echo -e "\033[1;31m==> ERROR:\033[1;0m \033[1;1m$1\033[1;0m" >&2
}

CONF=/etc/pacget.conf
STATS=/etc/pacget.stats
ARIA2=$(which aria2c 2> /dev/null)

# ----- do some checks first -----
if [ ! -x "$ARIA2" ]; then
  error "aria2c was not found or isn't executable."
  exit 1
fi

if [ $# -ne 2 ]; then
  error "Incorrect number of arguments"
  exit 1
fi

filename=$(basename $1)
server=${1%/$filename}
arch=$(grep ^Architecture /etc/pacman.conf | cut -d '=' -f2 | sed 's/ //g')
if [[ $arch = "auto" ]]; then
  arch=$(uname -m)
fi
# Determine which repo is being used
repo=$(awk -F'/' '$(NF-2)~/^(community|core|extra|testing|comunity-testing|multilib)$/{print $(NF-2)}' <<< $server)
[ -z $repo ] && repo="custom"

# For db files, or when using a custom repo (which most likely doesn't have any mirror),
# use only the URL passed by pacman; Otherwise, extract the list of servers (from the include file of the repo) to download from
url=$1
if ! [[ $filename = *.db || $repo = "custom" ]]; then
  mirrorlist=$(awk -F' *= *' '$0~"^\\["r"\\]",/Include *= */{l=$2} END{print l}' r=$repo /etc/pacman.conf)
  if [ -n mirrorlist ]; then
    num_conn=$(grep ^split $CONF | cut -d'=' -f2)
    url=$(sed -r '/^Server *= */!d; s/Server *= *//; s/\$repo'"/$repo/"'; s/\$arch'"/$arch/; s/$/\/$filename/" $mirrorlist | head -n $(($num_conn *2)) )
  fi
fi

msg "Downloading $filename"
cd /var/cache/pacman/pkg/

touch $STATS

$ARIA2 --conf-path=$CONF --max-tries=1 --max-file-not-found=5 \
  --uri-selector=adaptive --server-stat-if=$STATS --server-stat-of=$STATS \
  --allow-overwrite=true --remote-time=true --log-level=error --summary-interval=0 \
  $url --out=${filename}.pacget && [ ! -f ${filename}.pacget.aria2 ] && mv ${filename}.pacget $2 && chmod 644 $2

exit $?

/etc/pacget.conf

# The log file
log=/var/log/pacget.log
# Number of servers to download from
split=5
# Minimum file size that justifies a split, i.e. concurrent download (default 20M)
min-split-size=1M
# Maximum download speed (0 = unrestricted)
max-download-limit=0
# Minimum download speed (0 = do not care)
lowest-speed-limit=0
# Server timeout period
timeout=5
# 'none' or 'falloc'
file-allocation=none

Save this script as /usr/bin/pacget.

# chmod 755 /usr/bin/pacget

This makes the script an executable

In /etc/pacman.conf, in the [options] section, the following needs to be added:

XferCommand = exec /usr/bin/pacget %u %o

Using other applications

There are other downloading applications that you can use with Pacman. Here they are, and their associated XferCommand settings:

• snarf: XferCommand = /usr/bin/snarf -N %u
• lftp: XferCommand = /usr/bin/lftp -c pget %u
• axel: XferCommand = /usr/bin/axel -n 2 -v -a -o %o %u

Choosing the fastest mirror

When downloading packages pacman uses the mirrors in the order they are in /etc/pacman.d/mirrorlist. The mirror which is at the top of the list by default however may not be the fastest for you. To select a faster mirror, see Mirrors.

Sharing packages over your LAN

If you happen to run several Arch boxes on your LAN, you can share packages so that you can greatly decrease your download times. Keep in mind you should not share between different architectures (i.e. i686 and x86_64) or you'll get into troubles.

See Pacman_Tips#Network_shared_pacman_cache.