☰

壊れたインターネット接続を記録する方法

tag-icon • tag-icon ip monitoring ping
壊れたインターネット接続を記録する方法

ラズベリーパイがインターネットに接続したらすぐにスクリプトを実行する必要があります。しかし、毎分Googleをpingするよりも良い方法があるかどうか疑問に思います。

私の問題は、1日間インターネット接続が1〜2回切断されることです。したがって、そのようなイベントを記録する方法が必要です。
これは日中にADSLが落ちた場合に過ぎず、私が気づいていなくても、このようなことが起こったことを記録する方法を探していました。提案どおりにスクリプトを設定します。

答え1

あなたは確認できます：

cat /sys/class/net/wlan0/carrier

wlan0は私のインターネットインターフェースです。 eth0、eth1、wlan0など、使用しているインターフェイスを使用してインターネットに接続できます。このコマンドの出力が1の場合は接続されます。それ以外の場合は動作しません。したがって、次のスクリプトを作成できます。

#!/bin/bash
# Test for network conection
for interface in $(ls /sys/class/net/ | grep -v lo);
do
if [[ $(cat /sys/class/net/$interface/carrier) = 1 ]]; then ; echo "online"; fi
done

次のコマンドを使用することもできます。

#hwdetect --show-net

このスクリプトもよく実行されます。

#!/bin/bash

WGET="/usr/bin/wget"

$WGET -q --tries=20 --timeout=10 http://www.google.com -O   /tmp/google.idx &> /dev/null
if [ ! -s /tmp/google.idx ]
then
  echo "Not Connected..!"
else
  echo "Connected..!"
fi

答え2

システムに接続があると思うかどうかを尋ねるのは、「私のISPがデータを削除していますか？」のプロキシ測定です。定義によると、代理測定値は関心のあるシステムの単純化されたモデルであり、情報を保持しません。この質問は、興味のある情報を実際に得なければ答えを得ることができません。

ping実際にはしばしば特別な扱いを受けるICMPプロトコルなので、テストするのに悪い選択です。たとえば、HTTP接続に興味がある場合

curl --head http://www.example.com

実際にページを取得できるかどうかを示します。アンケートを行う場合は、慎重に行動し、通話の間に少なくとも60秒の睡眠時間を許可してください。 1分未満のISPの中断は「停止」と見なすことができます。

答え3

要件が比較的単純であることを考慮すると、単純なpingで十分です。 Google をテスト ホストとして使用する必要はありません。あなたのISPに公開ウェブサイトがある場合（最近はそうです）、そのウェブサイトを使用してください。

これは、古い、一時停止したプロジェクトに使用したい[bash]スクリプトです。単にpingホストとして機能し、統計を収集して必要に応じてシステムログに印刷します。唯一の依存関係pingはbc。あまりにも古いバージョンのpingLinux-Capability-awareが必要であることに注意してくださいping。

スクリプトは事前定義された間隔と署名値を使用し、対話式またはバックグラウンドで実行するように設計されています。たとえば、いくつかのロギングが必要な場合は、外部コマンドを提供する必要があります。同じ価格でジッタ検出とリンクアップ/ダウン検出機能を使用できます;-)。必要な場合に備えて助けが組み込まれています。ご不明な点がございましたら、お気軽にお問い合わせください。

私はメモリ使用量をできるだけ低く保ちながら、練習としてこれを書きました。ダッシュで実行させようと計画を立てた記憶が出ていますが、今まではこだわりを捨てることはできません。この内容が役立つことを願っています。内容は次のとおりです。

#!/bin/bash
#
#  ping.sh
#
#  Copyright 2014 VinzC <[email protected]>
#
#  This program is free software; you can redistribute it and/or modify
#  it under the terms of the GNU General Public License as published by
#  the Free Software Foundation; either version 2 of the License, or
#  (at your option) any later version.
#
#  This program is distributed in the hope that it will be useful,
#  but WITHOUT ANY WARRANTY; without even the implied warranty of
#  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#  GNU General Public License for more details.
#
#  You should have received a copy of the GNU General Public License
#  along with this program; if not, write to the Free Software
#  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
#  MA 02110-1301, USA.
#
# Runs in the background ping'ing given addresses. May send
# notifications through an external routine that is given as
# an argument.
#
# Uses environment variables:
#   * LOG_FILE (log file for debug purposes, silent otherwise)
#
# Symlinked as /usr/bin/anemon-ping

VERSION=1.0.3
AUTHOR="VinzC <[email protected]>"
DATE="Jan 2014"

# Background ping, keep only relevant information from the
# response, add timestamp in seconds. For use with graphing.

# Default padding for ICMP packets
ICMP_PADDING_DEFAULT=ffeab16b00b1e2

# Default number of seconds between ICMP probes
ICMP_INTERVAL_DEFAULT=300

# Number of samples to declare a host state steady
ICMP_STEADY_THRESHOLD=3

# Number of samples in which a host is allowed to go down
# 3 times before being flagged as flapping. (Why 3? 2 is
# not enough to make it a habbit, 3 is...)
ICMP_FLAPPING_THRESHOLD=24

# Number of samples after which dupe and damaged packets
# are gone for good. Default matches 24 hours.
ICMP_NET_ERROR_THRESHOLD=288

# Host state change command. The command is called with two or
# three arguments: host name and state, plus optional arguments.
# See usage for events.
host_cmd=false

# True if results shall be written to the standard output
verbose=false

# State variables
host_state=1            # Bit0: 1=up, 0=down
                        # Bit1: 1=flapping, 0=normal
host_unavail_state=0    # State counters
host_flap_state=0
host_warning_state=0

rtt_min=                # Ping statistics
rtt_max=
rtt_avg=

icmp_seq=0              # Number of requests
icmp_lost=0             # Count of losses

icmp_dupes=             # Network errors (counters)
icmp_damaged=

usage()
{
    cat <<EOF

USAGE
    ${0##*/} [-i interval] [-p pattern] [-P string] [-I interface]
        [-s packetsize] [-W timeout] [-v] [--host-command=CMD]
        [--flapping-threshold=N] [--error-threshold=N] destination
    ${0##*/} -h

ARGUMENTS
    Short options are a subset of standard ping arguments with the
    exception of -P and -v. Both -p and -P define a custom padding.
    Unlike ping's -p, option -P accepts any alphanumeric string,
    which will be converted to a hexadecimal string. The resulting
    pattern is eventually truncated to 16 bytes.

    Option -v switches to verbose mode. In this mode echo replies
    are sent to the standard output. This is useful for testing.

    -h  prints this help page.

    --host-command=CMD

        Run CMD on events. The first argument is the destination
        parameter, the second is the event name. Some events may
        have additional parameters.

        All events except the echo reply are sent to the system
        log. Events are:

        "start"     The monitoring process has been started. The
                external script can initialize its working
                context, e.g. create a round-robin database
                to store echo reply TTL values. An additional
                argument is passed with the interval value if
                option -i was specified. The default interval
                matches RRD default step, i.e. 300 seconds.
        "stop"      The monitoring process is stopping.
        "icmp"      Echo reply. Additional arguments are timestamp
                and roudtrip time. This event occurs every 5
                minutes by default and can be changed with -i.
        "up"        Host is up, steadily.
        "down"      Host is down, steadily.
        "flapping"  Host state is unstable.
        "damaged"   Damaged packets were detected.
        "dupes"     Duplicate packets were detected.

    --flapping-threshold=N

        The script attempts to detect flapping interfaces. It uses
        a discrete formula to keep a reasonable maximum delay in
        cases flapping occurs frequently, which must be considered
        a critical situation. Flapping detection is done as soon
        as no response from the remote host is received. No echo
        reply for more than 3 samples is a steady "down" state.

        An interface that has been flapping will take longer to be
        considered steady again. The value of --flapping-threshold
        sets that delay to the same amount of samples.

        The default value for N is 24. A bigger value will make an
        flapping interface wait proportionally longer before it is
        marked "up" again.

    --error-threshold=N

        Linux only: the background ping process informs whenever
        duplicated or damaged packets are received from a remote
        host. Such network errors are notified no more than once
        in N samples.

INTERACTIVE CONTROL
    The script reacts to signal SIGHUP to print statistics on the
    PING process running in the background. If verbose mode is
    enabled (-v) statistics are sent to the standard output. They
    are sent to the system log otherwise.

    Example:
        ${0##*/} --host-command=... host &
        PID=$!
        ...
        kill -HUP $PID

    PING statistics include min, avg and max roundtrip times and
    percentage of lost echo replies. Individual figures are also
    shown for damaged and duplicate packets if there are any.

    Note that the average value is a running average that uses a
    discrete averaging algorithm, i.e.:

        avg(i) = [ avg(i-1) + x(i-1) ] / 2

SEE ALSO
    man ping, ping(8)
EOF
    exit 0
}

help()
{
    cat <<EOF
${0##*/} version $VERSION, © $DATE by $AUTHOR

Monitors a remote host periodically sending ICMP packets from a ping
process running in the background. The script can execute a custom
command each time an important network condition occurs. The main
purpose is to record echo reply times in a round-robin database for
further graphing.
EOF
    usage
}

bc()
{
    # Workaround to print the leading zero for values <1
    /usr/bin/bc | sed 's/^\./0./g'
}

# Parse command line arguments. Only parse new or overriden arguments.
# Used to determine the remote host, mainly.
parse_args()
{
    # Need a F@!#^}G temporary variable to check getopt return code!
    args="$(getopt \
        -o i:I:p:P:s:W:hv \
        -l host-command:,flapping-threshold:,error-threshold \
        -- "$@")" && eval set -- "$args" && unset args || return $?

    # Now check the remaining arguments
    while [ -n "$1" ]; do
        [ "$1" = "--" ] || case $1 in

            --host-command)
                host_cmd=$2; shift;;

            --flapping-threshold)
                ICMP_FLAPPING_THRESHOLD=$2; shift;;

            --error-threshold)
                ICMP_NET_ERROR_THRESHOLD=$2; shift;;

            -i)
                ICMP_INTERVAL=$2; shift;;
            -p)
                ICMP_PADDING=$2; shift;;
            -P)
                ICMP_PADDING=$(printf "$2" | od -A n -t x1 | \
                    sed -r -e 's:\s+::g' -e 's:.::33g'); shift;;
            -I)
                ICMP_IFACE=$2; shift;;
            -s)
                ICMP_PKTSIZE=$2; shift;;
            -W)
                ICMP_TIMEOUT=$2; shift;;
            -v)
                verbose=true;;
            -h)
                help;;
            *)
                ICMP_HOST="$1";;
        esac; shift
    done
    [ -n "$ICMP_HOST" ] || usage 1>&2
}

logger()
{
    # Write to standard output in verbose mode, to syslog otherwise
    $verbose && echo "$@" || \
        /usr/bin/logger -t "${0##*/}[$$]" "$@"
}

set_response_time()
{
    # Call external command or prompt to the console in verbose mode
    $host_cmd $ICMP_HOST "icmp" $1 $3 && ! $verbose || \
        printf "%d: seq=%d, time=%s\n" "$1" "$2" "$3"
    return 0
}

set_state_up()
{
    # Clear flapping state ans set (steady) up flag
    host_state=0x01

    # Call the external notification function and log host state
    $host_cmd $ICMP_HOST "up"
    logger "Host interface or host @ $ICMP_HOST is now up."
}

set_state_down()
{
    # Clear up flag only
    host_state=$(( host_state & 0xFE ))

    # Call the external notification function and log host state
    $host_cmd $ICMP_HOST "down"
    logger "Host interface or host @ $ICMP_HOST is down!"
}

set_state_flapping()
{
    # Set flapping and down flags
    host_state=2

    # Call the external notification function and log host state
    $host_cmd $ICMP_HOST "flapping"
    logger "Host interface or host @ $ICMP_HOST is unstable!"
}

set_host_message()
{
    # Reset error counter to the maximum
    host_warning_state=$ICMP_NET_ERROR_THRESHOLD

    # Call the external notification function and log host state
    $host_cmd $ICMP_HOST $1
    logger "Errors received from host interface or host @ $ICMP_HOST ($1)!"
}

print_stats()
{
    if [ $icmp_seq -eq 0 ]; then
        logger "PING $ICMP_HOST: no packet sent"
    else
        local icmp_received=$(( icmp_seq - icmp_lost ))
        local icmp_losses=$( echo "scale=2; 100 * $icmp_lost / $icmp_seq" | bc )
        logger "PING $ICMP_HOST: $icmp_seq packets sent, $icmp_lost lost, ${icmp_losses}% loss${rtt_min:+; rtt min/avg/max = $rtt_min/$rtt_avg/$rtt_max ms}${icmp_dupes:+, $icmp_dupes dupes}${icmp_damaged:+, $icmp_damaged bad CRC}"
    fi
}

echo_reply()
{
    # First argument is time in seconds (icmp_seq is global)
    local TM=$1 ttl time msg; shift

    # Evaluate the remaining arguments as expressions
    eval "$@"

    # No time variable means host is not responding
    [ -z "$time" ] && return 1

    # Update statistics: average, minimum and maximum RTT
    rtt_avg=$( echo "scale=3; (${rtt_avg:-$time} + $time)/2" | bc )
    rtt_min=$( echo "scale=3; rtt_min=${rtt_min:-$time}; if ($time < rtt_min) $time else rtt_min" | bc )
    rtt_max=$( echo "scale=3; rtt_max=${rtt_max:-$time}; if ($time > rtt_max) $time else rtt_max" | bc )

    # Decrement the state counter if greater than zero
    [ $host_unavail_state -ne 0 ] && \
        host_unavail_state=$(( host_unavail_state - 1 ))

    # The host is not up if:
    # - it is flapping (bit 1 of the state flag) and the flapping
    #   counter is greater than 0  OR
    # - the state counter is greater than 0 (non flapping case).
    # As long as one of these condition is true, the state flag
    # will not be set to UP (bit 0 set, bit 1 cleared).
    #
    # Once the state counter reaches zero (steady "up" state) and
    # the host state is no longer flapping, change the state flag.
    # Also don't change the state flag if bit 0 was already set.
    [ $(( host_state & 0x02 )) -ne 0  ] && [ $host_flap_state -ne 0 ] || \
    [ $host_unavail_state -ne 0 ] || \
    [ $(( host_state & 0x01 )) -ne 0 ] || \
        set_state_up

    # Warn if damaged or duplicate packets. Don't warn
    # again until the warning counter reaches zero. Treat
    # damaged and dupe packets alike for both are very
    # unlikely to occur at the same time.
    if [ -n "$msg" ]; then
        eval rtt_$msg=$(( rtt_$msg + 1 ))
        [ $host_warning_state -eq 0 ] && set_host_message $msg
    fi

    # Run external command to store response time
    set_response_time $TM $icmp_seq $time
    return 0
}

no_response()
{
    # Store the number of lost replies
    icmp_lost=$(( icmp_lost + 1 ))

    # FLAPPING DETECTION
    # ------------------
    # Increment flapping state using a discrete low-pass formula
    # to prevent excessive values. Handle flapping only if host
    # has just come down, don't wait for a steady "down" state.
    [ $host_unavail_state -eq 0 ] && \
    [ $(( host_flap_state=(3*host_flap_state + 7*ICMP_FLAPPING_THRESHOLD) / 8 )) -gt $ICMP_FLAPPING_THRESHOLD ] && \
        set_state_flapping

    # Increment host state until it reaches the threshold, which
    # marks the steady "down" state. Only then call the external
    # command to allow notifying the host is "down". Just don't
    # call the command more than once if the host is still down
    # by the next time.
    [ $host_unavail_state -lt $ICMP_STEADY_THRESHOLD ] && \
    [ $(( host_unavail_state=host_unavail_state + 1 )) -eq $ICMP_STEADY_THRESHOLD ] && \
    [ $(( host_state & 0x03 )) -eq 1 ] && \
        set_state_down
}

# Parse command-line arguments and set globals
parse_args "$@" || exit $?

# Redirect stderr to LOG_FILE if defined
[ -z "$LOG_FILE" ] || exec 2>$LOG_FILE

# Print PING statistics upon receiving SIGUSR1
trap print_stats HUP

# Send even "stop" upon terminating
trap "printf '\n'; $host_cmd $ICMP_HOST stop; print_stats" INT QUIT TERM ABRT


# Notify monitoring starts
$host_cmd $ICMP_HOST "start"

# 1. filter out lines keeping only those that include response
#    times and those about non responding hosts.
# 2. Stick units to response times and keep only the multiplier
#    if it's different from "m"
# 3. Keep only the integer part of the timestamp, erase garbage
#    before the relevant information (var=value)
# 4. Warn about damaged and duplicate packets
#
# Make sure sed does NOT buffer output (hence -u)
while read R; do
    echo_reply $R || no_response

    # Decrement other state variables until it reaches zero
    [ $host_flap_state -eq 0 ] || \
        host_flap_state=$(( host_flap_state - 1 ))

    [ $host_warning_state -eq 0 ] || \
        host_warning_state=$(( host_warning_state - 1 ))

# Downside: need bash for process redirection, which is needed
# to access state variables outside the while/read loop...
done < <(LC_ALL=POSIX /bin/ping -OD \
    ${ICMP_IFACE:+-I $ICMP_IFACE} \
    ${ICMP_TIMEOUT:+-W $ICMP_TIMEOUT} \
    ${ICMP_PKTSIZE:+-s $ICMP_PKTSIZE} \
    -i ${ICMP_INTERVAL:-$ICMP_INTERVAL_DEFAULT} \
    -p ${ICMP_PADDING:-$ICMP_PADDING_DEFAULT} $ICMP_HOST |
sed -rnu \
    -e '/no answer|[0-9]+ bytes from/!d' \
    -e 's@(time=[0-9.]+)\s+m?(\w*)s@\1\2@g' \
    -e 's@\(DUP\!\)@msg="dupes"@g' \
    -e 's@\(BAD CHECKSUM\!\)@msg="damaged"@g' \
    -e 's@\[(\w+)\.\w+\][a-zA-Z0-9():. \-]+\s+@\1 @gp')

答え4

ここに説明されているシェルスクリプトをインストールします。 クーポンシェルスクリプト。<ad>これはpingがより簡単になり、スマートになりました。。</ad>

インストールされたら、ISP（たとえば、最上位の接続に必要なDNSサーバーの1つを使用）とISPの背後にあるインターネット（Googleサーバーなど）。

これを並行して実行すると、ISP アクセスまたはインターネットに問題があるかどうかがわかります。

関連情報