X-Git-Url: https://iankelling.org/git/?a=blobdiff_plain;f=system-status;h=c11024bc5ba2c125615beb78c285f6193755c201;hb=802e885e3e7fa3857f8bc4f54c261d5ca76f2454;hp=051de623df85e046d22a01f7d810b2e597f42578;hpb=d4366929e6e200155b010dc05ce74255ee6a45ed;p=distro-setup diff --git a/system-status b/system-status old mode 100644 new mode 100755 index 051de62..c11024b --- a/system-status +++ b/system-status @@ -2,11 +2,16 @@ # Copyright (C) 2019 Ian Kelling # SPDX-License-Identifier: AGPL-3.0-or-later -# usage: runs 4 times every 15 seconds unless any args are passed, or we -# are on battery power, then just runs once. +# usage: runs once every 15 seconds unless any args are passed, or we +# then just runs once. On battery power, run once per minute. if [ -z "$BASH_VERSION" ]; then echo "error: shell is not bash" >&2; exit 1; fi +if [[ $EUID != 1000 ]]; then + echo "$0: error, expected to be user 1000" + exit 1 +fi + source /a/bin/errhandle/err status_file=/dev/shm/iank-status @@ -28,31 +33,152 @@ v() { printf "%s\n" "$*" fi } +p() { printf "%s\n" "$*"; } # log-once COUNT NAME [MESSAGE] lo() { - /usr/local/bin/log-once "$@" | ifne mail -s "$HOSTNAME: system-status $2" root@localhost + if type -p ifne &>/dev/null; then + /usr/local/bin/log-once "$@" | ifne mail -s "$HOSTNAME: system-status $2" root@localhost + fi +} + +loday() { + if type -p ifne &>/dev/null; then + /usr/local/bin/log-once "$@" | ifne mail -s "$HOSTNAME: system-status $2" daylert@iankelling.org + fi } +# todo, consider migrating some of these alerts into prometheus write-status() { chars=("${first_chars[@]}") - # clock us out in timetrap if are idle too long - if [[ -e /p/.timetrap.db ]]; then - export DISPLAY=:0 - if type -p xprintidle &>/dev/null && xidle=$(xprintidle 2>/dev/null); then - if [[ $xidle == [0-9]* ]]; then - sheet=$(sqlite3 /p/.timetrap.db "select sheet from entries where end is NULL;") - idle=300000 - if [[ $sheet == w ]]; then - idle=900000 - fi - if [[ $sheet && $xidle -gt $idle ]]; then - timetrap out - fi + services=( epanicclean ) + case $HOSTNAME in + bk|je|li) : ;; + *) + services+=( + systemstatus + btrfsmaintstop + dynamicipupdate + ) + bads=() + if systemctl show -p SubState --value ${services[@]} | egrep -v '^(running|)$' &>/dev/null; then + for s in ${services[@]}; do + if [[ $(systemctl show -p SubState --value $s 2>&1) != running ]]; then + bads+=($s) + fi + done + chars+=(MYSERS) fi - fi + p ${bads[*]} | lo -240 mysers + ;; + esac + + case $HOSTNAME in + kd) + services=( + prometheus-node-exporter + prometheus-alertmanager + prometheus + ) + bads=() + if systemctl show -p SubState --value ${services[@]} | egrep -v '^(running|)$' &>/dev/null; then + for s in ${services[@]}; do + if [[ $(systemctl show -p SubState --value $s 2>&1) != running ]]; then + bads+=($s) + fi + done + chars+=(PROM) + fi + p ${bads[*]} | lo -240 prom + ;; + esac + + + if [[ -e /a/bin/bash_unpublished/source-state ]]; then + # /a gets remounted due to btrbk, ignore error code for file doesnt exist + source /a/bin/bash_unpublished/source-state || [[ $? == 1 ]] fi + if [[ $MAIL_HOST == "$HOSTNAME" ]]; then + + bouncemsg= + glob=(/m/md/bounces/new/*) + if [[ -e ${glob[0]} ]]; then + chars+=(BOUNCE) + bouncemsg="message in /m/md/bounces/new" + fi + p $bouncemsg | loday -1 bounce + # emails without the S (seen) flag. this only checks the last flag, + # but its good enough for me. + glob=(/m/md/alerts/{new,cur}/!(*,S)) + if [[ -e ${glob[0]} ]]; then + chars+=(A) + fi + + glob=(/m/md/daylert/{new,cur}/!(*,S)) + if [[ -e ${glob[0]} ]]; then + chars+=(DAY) + fi + + bbkmsg= + if [[ $(systemctl is-active btrbk.timer) != active ]]; then + chars+=(BTRBK.TIMER) + bbkmsg="not enabled" + fi + p "$bbkmsg" | lo -480 btrbk.timer + + ## check if last snapshot was within an hour + vol=o + # this section generally copied from btrbk scripts, but + # this part modified to speed things up by about half a second. + # I'm not sure if its quite as reliable, but it looks pretty safe. + # Profiled it using time and also adding to the top of the file: + # set -x + # PS4='+ $(date "+%2N") ' + # allow failure in case there are no snapshots yet. + # shellcheck disable=SC2012 + shopt -u nullglob + files=(/mnt/root/btrbk/$vol.20*) + shopt -s nullglob + snaps=() + if (( ${#files[@]} )); then + snaps=($(ls -1avdr "${files[@]}" 2>/dev/null |head -n1 || : )) + fi + now=$(date +%s) + maxtime=0 + for s in ${snaps[@]}; do + file=${s##*/} + t=$(date -d $(sed -r 's/(.{4})(..)(.{5})(..)(.*)/\1-\2-\3:\4:\5/' <<<${file#$vol.}) +%s) + if (( t > maxtime )); then + maxtime=$t + fi + done + snapshotmsg= + if (( maxtime < now - 4*60*60 )); then + chars+=(OLD-SNAP) + snapshotmsg="/o snapshot older than 4 hours" + fi + p "$snapshotmsg" | lo -1 old-snapshot + + + # commented out, only using timetrap retrospectively. + # # clock us out in timetrap if are idle too long + # if [[ -e /p/.timetrap.db ]]; then + # export DISPLAY=:0 + # if type -p xprintidle &>/dev/null && xidle=$(xprintidle 2>/dev/null); then + # if [[ $xidle == [0-9]* ]]; then + # sheet=$(sqlite3 /p/.timetrap.db "select sheet from entries where end is NULL;") + # idle=300000 + # if [[ $sheet == w ]]; then + # idle=900000 + # fi + # if [[ $sheet && $xidle -gt $idle ]]; then + # timetrap out + # fi + # fi + # fi + # fi + fi if ip l show tunfsf &>/dev/null; then # this is for tracking dns over tls issue, which @@ -78,35 +204,28 @@ write-status() { glob=(/nocow/btrfs-stale/*) if [[ -e ${glob[0]} ]]; then - chars+=("STALE") + chars+=(STALE) fi + var_mail_msg= if [[ $(find /var/mail -type f \! -empty -print -quit) ]]; then var_mail_msg="message in /var/mail" fi - lo -1 var_mail $var_mail_msg - glob=(/m/md/bounces/new/*) - if [[ -e ${glob[0]} ]]; then - chars+=("BOUNCE") - bouncemsg="message in /m/md/bounces/new" - fi - lo -1 bounce $bouncemsg - # emails without the S (seen) flag. this only checks the last flag, - # but its good enough for me. - glob=(/m/md/alerts/{new,cur}/!(*,S)) - if [[ -e ${glob[0]} ]]; then - chars+=("A") - fi + p $var_mail_msg | loday -1 var_mail + + tmp=(/var/local/cron-errors/mailtest-check*) if (( ${#tmp[@]} )); then - chars+=("MAILPING") + chars+=(MAILPING) fi tmp=(/var/local/cron-errors/mailtest-slow*) if (( ${#tmp[@]} )); then - chars+=("SPAMD") + chars+=(SPAMD) fi - # early in install process, we dont have permission yet for exiqgrep - qlen=$(/usr/sbin/exiqgrep -o 600 -c -b | awk '{print $1}') ||: + # early in install process, we dont have permission yet for exiqgrep. + # 1100 helps allow for system restarts + qlen=$(/usr/sbin/exiqgrep -o 1100 -c -b | awk '{print $1}') ||: + qmsg= if ((qlen)); then qmsg="queue length $qlen" chars+=("q $qlen") @@ -114,28 +233,29 @@ write-status() { case $HOSTNAME in # No point in emailing about the mailq on a host where we don't # check email. - $MAIL_HOST|bk) - lo -120 qlen $qmsg + $MAIL_HOST) + p $qmsg | loday -120 qlen ;; esac begin=false - if ! make -C /b/ds -q ~/.local/distro-begin || [[ $(<~/.local/distro-begin) != 0 ]]; then + + if ! make -C /b/ds -q ~/.local/distro-begin 2>/dev/null || [[ $(<~/.local/distro-begin) != 0 ]]; then begin=true fi end=false - if ! make -C /b/ds -q ~/.local/distro-end || [[ $(<~/.local/distro-end) != 0 ]]; then + if ! make -C /b/ds -q ~/.local/distro-end 2>/dev/null || [[ $(<~/.local/distro-end) != 0 ]]; then end=true fi # these conditions are so we dont have an overly verbose prompt if $begin && $end; then - chars+=("D") + chars+=(D) elif $begin; then - chars+=("DB") + chars+=(DB) elif $end; then - chars+=("DE") + chars+=(DE) else f=~/.local/conflink # shellcheck disable=SC2043 @@ -162,7 +282,7 @@ write-status() { # Just because i forget a lot, -mmin -NUM means files modified <= NUM minutes ago if (( fmin < 0 )) && [[ $(find ${all_dirs[@]} -mmin $fmin -type f -print -quit 2>/dev/null) ]]; then v conflink newer filesystem files - chars+=("CONFLINK") + chars+=(CONFLINK) break fi @@ -175,7 +295,7 @@ write-status() { fi if (( $(date -d "$(git log --diff-filter=ACR --format=%aD -1)" +%s) > fsec )); then v conflink: newer files checked in to git - chars+=("CONFLINK") + chars+=(CONFLINK) break fi @@ -185,7 +305,7 @@ write-status() { done < <(git ls-files -o --exclude-standard) if [[ ${untracked[0]} && $(find "${untracked[@]}" -mmin $fminplus -type f -print -quit) ]]; then v conflink: untracked in $d - chars+=("CONFLINK") + chars+=(CONFLINK) break fi done @@ -194,74 +314,31 @@ write-status() { fi if [[ ! -e $f || $(<$f) != 0 ]]; then v conflink: last run not found or failed - chars+=("CONFLINK") + chars+=(CONFLINK) break fi done fi -# if [[ $(grep -v "exim user lost privilege for using -C option" /var/log/exim4/paniclog 2>/dev/null ||:) ]]; then + # if [[ $(grep -v "exim user lost privilege for using -C option" /var/log/exim4/paniclog 2>/dev/null ||:) ]]; then if [[ -s /var/log/exim4/paniclog ]]; then chars+=("PANIC!") # leave it up to epanic-clean to send email notification fi - source /a/bin/bash_unpublished/source-state - if [[ $MAIL_HOST == "$HOSTNAME" ]]; then - bbkmsg= - if [[ $(systemctl is-active btrbk.timer) != active ]]; then - chars+=("BTRBK.TIMER") - bbkmsg="btrbk.timer not enabled" + if [[ ! -e $status_file || -w $status_file ]]; then + if [[ -e /a/bin/bash_unpublished/source-state ]]; then + cat /a/bin/bash_unpublished/source-state >$status_file fi - lo -960 btrbk.timer $bbkmsg - ## check if last snapshot was within an hour - vol=o - # this section generally copied from btrbk scripts, but - # this part modified to speed things up by about half a second. - # I'm not sure if its quite as reliable, but it looks pretty safe. - # Profiled it using time and also adding to the top of the file: - # set -x - # PS4='+ $(date "+%2N") ' - # allow failure in case there are no snapshots yet. - # shellcheck disable=SC2012 - shopt -u nullglob - files=(/mnt/root/btrbk/$vol.20*) - shopt -s nullglob - snaps=() - if (( ${#files[@]} )); then - snaps=($(ls -1avdr "${files[@]}" 2>/dev/null |head -n1 || : )) - fi - now=$(date +%s) - maxtime=0 - for s in ${snaps[@]}; do - file=${s##*/} - t=$(date -d $(sed -r 's/(.{4})(..)(.{5})(..)(.*)/\1-\2-\3:\4:\5/' <<<${file#$vol.}) +%s) - if (( t > maxtime )); then - maxtime=$t - fi - done - if (( maxtime < now - 4*60*60 )); then - chars+=("OLD-SNAP") - snapshotmsg="/o snapshot older than 4 hours" + if [[ ${chars[*]} ]]; then + echo "ps_char=\"${chars[*]} \$ps_char\"" >>$status_file fi - lo -1 old-snapshot $snapshotmsg fi - - cat /a/bin/bash_unpublished/source-state >$status_file - - if [[ ${chars[*]} ]]; then - echo "ps_char=\"${chars[*]} \$ps_char\"" >>$status_file - fi - } # use this if we want to do something just once per minute first_chars=() -power=true -if [[ -e /sys/class/power_supply/AC/online && $(