X-Git-Url: https://iankelling.org/git/?p=distro-setup;a=blobdiff_plain;f=system-status;h=c11024bc5ba2c125615beb78c285f6193755c201;hp=f50d238b7d6523acc14c65765503894a52811916;hb=802e885e3e7fa3857f8bc4f54c261d5ca76f2454;hpb=40dd151ec6ba75633c74568da59e35a45351f194 diff --git a/system-status b/system-status index f50d238..c11024b 100755 --- a/system-status +++ b/system-status @@ -7,6 +7,11 @@ if [ -z "$BASH_VERSION" ]; then echo "error: shell is not bash" >&2; exit 1; fi +if [[ $EUID != 1000 ]]; then + echo "$0: error, expected to be user 1000" + exit 1 +fi + source /a/bin/errhandle/err status_file=/dev/shm/iank-status @@ -28,6 +33,7 @@ v() { printf "%s\n" "$*" fi } +p() { printf "%s\n" "$*"; } # log-once COUNT NAME [MESSAGE] lo() { if type -p ifne &>/dev/null; then @@ -45,7 +51,6 @@ loday() { write-status() { chars=("${first_chars[@]}") - services=( epanicclean ) case $HOSTNAME in bk|je|li) : ;; @@ -55,28 +60,26 @@ write-status() { 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 + p ${bads[*]} | lo -240 mysers ;; esac - 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 - lo -240 mysers ${bads[*]} - - services=( - prometheus-node-exporter - prometheus-alertmanager - prometheus - ) 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 @@ -86,27 +89,96 @@ write-status() { done chars+=(PROM) fi - lo -240 prom ${bads[*]} + p ${bads[*]} | lo -240 prom ;; esac - # clock us out in timetrap if are idle too long - if [[ -e /p/.timetrap.db ]]; then - export DISPLAY=:0 - if type -p xprintidle &>/dev/null && xidle=$(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 + + 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 - 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 @@ -138,26 +210,7 @@ write-status() { if [[ $(find /var/mail -type f \! -empty -print -quit) ]]; then var_mail_msg="message in /var/mail" fi - loday -1 var_mail $var_mail_msg - - bouncemsg= - glob=(/m/md/bounces/new/*) - if [[ -e ${glob[0]} ]]; then - chars+=(BOUNCE) - bouncemsg="message in /m/md/bounces/new" - fi - loday -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 - - glob=(/m/md/daylert/{new,cur}/!(*,S)) - if [[ -e ${glob[0]} ]]; then - chars+=(DAY) - fi + p $var_mail_msg | loday -1 var_mail tmp=(/var/local/cron-errors/mailtest-check*) @@ -180,8 +233,8 @@ write-status() { case $HOSTNAME in # No point in emailing about the mailq on a host where we don't # check email. - $MAIL_HOST|bk) - loday -120 qlen $qmsg + $MAIL_HOST) + p $qmsg | loday -120 qlen ;; esac @@ -273,52 +326,6 @@ write-status() { # leave it up to epanic-clean to send email notification fi - 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 - bbkmsg= - if [[ $(systemctl is-active btrbk.timer) != active ]]; then - chars+=(BTRBK.TIMER) - bbkmsg="not enabled" - fi - lo -480 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 - snapshotmsg= - if (( maxtime < now - 4*60*60 )); then - chars+=(OLD-SNAP) - snapshotmsg="/o snapshot older than 4 hours" - fi - lo -1 old-snapshot $snapshotmsg - fi - 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 @@ -328,7 +335,6 @@ write-status() { echo "ps_char=\"${chars[*]} \$ps_char\"" >>$status_file fi fi - } # use this if we want to do something just once per minute first_chars=()