#!/bin/bash
-# Usage: mail-test-check [slow] [anything]
+# Usage: mail-test-check [slow] [int|nonint]
#
# slow: do slow checks, like spamassassin
#
-# anything: consider non-interactive, dont print unless something went
+# for non-interactive, dont print unless something went
# wrong
-source /b/errhandle/err
+source /b/bash-bear-trap/bash-bear
[[ $EUID == 0 ]] || exec sudo -E "${BASH_SOURCE[0]}" "$@"
e() { $int || return 0; printf "mailtest-check: %s\n" "$*"; }
-
-## Minutes before we give error.
-# We run this cronjob along with sending the test email every 5 minutes,
-# so give it 1 minute to arrive, then if the latest email is older than
-# 7 minutes, the last 2 haven't arrived in a reasonable amount of time.
-min_limit=8
+getspamdpid() {
+ if [[ ! $spamdpid || ! -d /proc/$spamdpid ]]; then
+ # try twice in case we are restarting, it happens.
+ for i in 1 2; do
+ spamdpid=$(systemctl show --property MainPID --value spamassassin | sed 's/^[10]$//' ||:)
+ if [[ $spamdpid ]]; then
+ break
+ fi
+ sleep 30
+ done
+ fi
+}
-# spamassassin checking takes about 8 seconds. only do that every
-# once in a while.
+#### begin arg processing ####
+# spamassassin checking takes about 8 seconds.
slow=false
if [[ $1 == slow ]]; then
slow=true
if [[ $1 == nonint ]]; then
int=false
fi
+#### end arg processing ####
+# we put this in to avoid dns errors that happen on reboot,
+# but I want to debug them.
+# if ! $int; then
+# sleep 60
+# fi
-if ! $int; then
- sleep 60
-fi
-# avoid errors like this:
-# Nov 8 08:16:05.439 [6080] warn: plugin: failed to parse plugin (from @INC): Can't locate Mail/SpamAssassin/Plugin/WLBLEval.pm: lib/Mail/SpamAssassin/Plugin/WLBLEval.pm: Permission denied at (eval 59) line 1.
-#Nov 8 08:16:05.439 [6080] warn: plugin: failed to parse plugin (from @INC): Can't locate Mail/SpamAssassin/Plugin/VBounce.pm: lib/Mail/SpamAssassin/Plugin/VBounce.pm: Permission denied at (eval 60) line 1.
-# i dont know why, i just found the solution online
-cd /m/md
+# TODO, get je to deliver the local mailbox: /m/md/INBOX
+# dovecot appears to setup, i can t be sure.
+
+source /a/bin/bash_unpublished/source-state
+doprom=false
case $HOSTNAME in
- bk)
- folders=(/m/md/{expertpathologyreview.com,amnimal.ninja}/testignore)
- froms=(ian@iankelling.org z@zroe.org testignore@je.b8.nz iank@gnu.org)
- ;;
- je)
- froms=(ian@iankelling.org z@zroe.org testignore@expertpathologyreview.com testignore@amnimal.ninja)
- folders=(/m/md/je.b8.nz/testignore)
+ $MAIL_HOST|bk|je)
+ doprom=true
;;
*)
- folders=(/m/md/l/testignore)
- froms=(testignore@je.b8.nz testignore@expertpathologyreview.com testignore@amnimal.ninja ian@iankelling.org z@zroe.org iank@gnu.org)
- if ! $int; then
- timeout 120 rsync --chown iank:iank -e "ssh -oIdentitiesOnly=yes -F /dev/null -i /root/.ssh/jtuttle" -t --inplace -r 'jtuttle@fencepost.gnu.org:/home/j/jtuttle/Maildir/new/' /m/md/l/testignore/new
- fi
+ rm -f /var/lib/prometheus/node-exporter/mailtest-check.prom*
;;
esac
-getspamdpid() {
- if [[ ! $spamdpid || ! -d /proc/$spamdpid ]]; then
- spamdpid=$(systemctl show --property MainPID --value spamassassin | sed 's/^1$//' ||:)
- fi
-}
-getspamdpid
-pr() {
- cat >>/var/lib/prometheus/node-exporter/mailtest-check.prom.$$
-}
-pr <<EOF
+main() {
+
+ local -a p_unexpected_spamd_results p_missing_dnswl p_last_usec
+ case $HOSTNAME in
+ bk)
+ folders=(/m/md/{expertpathologyreview.com,amnimal.ninja}/testignore)
+ froms=(ian@iankelling.org z@zroe.org testignore@je.b8.nz iank@gnu.org)
+ ;;
+ je)
+ froms=(ian@iankelling.org z@zroe.org iank@gnu.org testignore@amnimal.ninja)
+ folders=(/m/md/je.b8.nz/testignore)
+ ;;
+ *)
+ folders=(/m/md/l/testignore)
+ froms=(testignore@je.b8.nz testignore@expertpathologyreview.com testignore@amnimal.ninja ian@iankelling.org z@zroe.org)
+ if ! $int; then
+ ### begin rsyncing fencepost email ###
+ # We dont want to exit if rsync fails, that will get caught by
+ # our later test by virtue of not having the latest email.
+ did_rsync=false
+ try_start_time=$EPOCHSECONDS
+ try_limit=140 # somewhat arbitrary value
+ while ! $did_rsync; do
+ try_left=$(( try_limit - ( EPOCHSECONDS - try_start_time) ))
+ timeout=120 # somewhat arbitrary value
+ if (( try_left < 0 )); then
+ echo "mailtest-check: failed to rsync fencepost > $try_limit seconds"
+ break
+ fi
+ if (( try_left < timeout )); then
+ timeout=$try_left
+ fi
+ if timeout $timeout rsync --chown iank:iank -e "ssh -oIdentitiesOnly=yes -F /dev/null -i /root/.ssh/jtuttle" -t --inplace -r 'jtuttle@fencepost.gnu.org:/home/j/jtuttle/Maildir/new/' /m/md/l/testignore/new; then
+ did_rsync=true
+ else
+ sleep 4
+ fi
+ done
+ if ! $did_rsync; then
+ echo mailtest-check: warning: fencepost rsync failed
+ fi
+ ### end rsyncing fencepost email ###
+ fi
+ ;;
+ esac
+
+
+ # avoid errors like this:
+ # Nov 8 08:16:05.439 [6080] warn: plugin: failed to parse plugin (from @INC): Can't locate Mail/SpamAssassin/Plugin/WLBLEval.pm: lib/Mail/SpamAssassin/Plugin/WLBLEval.pm: Permission denied at (eval 59) line 1.
+ #Nov 8 08:16:05.439 [6080] warn: plugin: failed to parse plugin (from @INC): Can't locate Mail/SpamAssassin/Plugin/VBounce.pm: lib/Mail/SpamAssassin/Plugin/VBounce.pm: Permission denied at (eval 60) line 1.
+ # i dont know why, i just found the solution online
+ cd /m/md
+
+
+ getspamdpid
+ # first time we write, overwrite anything existing
+ if [[ -e /var/lib/prometheus/node-exporter ]]; then
+ cat >/var/lib/prometheus/node-exporter/mailtest-check.prom.$$ <<EOF
mailtest_check_found_spamd_pid_bool $(( ${spamdpid:-0} > 0 ))
EOF
-e spamdpid: $spamdpid
-if [[ ! $spamdpid ]]; then
- echo $HOSTNAME mailtest spamd pid not found. systemctl status spamassassin:
- systemctl status spamassassin
-fi
-tmpfile=$(mktemp)
-declare -i unexpected=0
-for folder in ${folders[@]}; do
- for from in ${froms[@]}; do
- latest=
- last_sec=0
-
- if ! grep -rlFx "From: $from" $folder/{new,cur} >$tmpfile; then
- e "no message found from: $from"
- continue
- fi
- # webmail sends them to cur it seems
- while read -r file; do
- if [[ $file -nt $latest ]]; then
- latest=$file
+ fi
+ e spamdpid: $spamdpid
+ if [[ ! $spamdpid ]]; then
+ echo mailtest spamd pid not found. systemctl status spamassassin:
+ systemctl status spamassassin
+ fi
+ tmpfile=$(mktemp)
+ declare -i unexpected=0
+ for folder in ${folders[@]}; do
+ for from in ${froms[@]}; do
+ declare -i missing_dnswl=0
+ #declare -i dnsfail=0
+ declare -i unexpected=0
+ latest=
+ last_sec=0
+
+ if ! grep -rlFx "From: $from" $folder/{new,cur} >$tmpfile; then
+ echo "no message found from: $from"
+ continue
fi
- done <$tmpfile
+ # webmail sends them to cur it seems
+ while read -r file; do
+ file_sec=$(awk '/^Subject: / {print $4}' $file)
+ if [[ $file_sec ]] && (( file_sec > last_sec )); then
+ latest=$file
+ last_sec="$file_sec"
+ fi
+ done <$tmpfile
+ rm -f $tmpfile
- if [[ ! $latest ]]; then
- # 10 is an arbitrary bad value
- unexpected+=10
- else
to=$(awk '/^Envelope-to: / {print $2}' $latest)
- last_sec=$(awk '/^Subject: / {print $4}' $latest)
if $slow; then
if ! $int; then
if [[ $(readlink /proc/$$/ns/net) != "$(readlink /proc/$spamdpid/ns/net)" ]]; then
spamcpre="nsenter -t $spamdpid -n -m"
fi
-
+ unset results
declare -A results
# pyzor fails for our test message, so dont put useless load on their
# servers.
# example line that sed is parsing:
# (-0.1 / 5.0 requ) DKIM_SIGNED=0.1,DKIM_VALID=-0.1,DKIM_VALID_AU=-0.1,SPF_HELO_PASS=-0.001,SPF_PASS=-0.001,TVD_SPACE_RATIO=0.001 autolearn=_AUTOLEARN
- for r in $($spamcpre sudo -u Debian-exim spamassassin -t --cf='score PYZOR_CHECK 0' <"$latest" | tail -n2 | head -n1 | sed -r 's/^\([^)]*\) *//;s/=[^, ]*([, ]|$)/ /g'); do
+ resultfile=$(mktemp)
+ $spamcpre sudo -u Debian-exim spamassassin -D -t --cf='score PYZOR_CHECK 0' <"$latest" &>$resultfile
+
+ # note: on some mail, its 1 line after the send-test-forward, on others its 2 with a blank inbetween.
+ # I use the sed -n to filter this.
+ raw_results="$(tail $resultfile | grep -A2 -Fx /usr/local/bin/send-test-forward | tail -n+2 | sed -nr 's/^\([^)]*\) *//;s/=[^, ]*([, ]|$)/ /gp')"
+ for r in $raw_results; do
case $r in
# got this in an update 2022-01. dun care
T_SCC_BODY_TEXT_LINE|SCC_BODY_SINGLE_WORD) : ;;
# it seems like some versions of spamassassin do BODY_SINGLE_WORD, others dont, we dun care.
# bayes_00 is a new one indicating ham, we dont care if its missing.
BAYES_00|BODY_SINGLE_WORD|FROM_FMBLA_NEWDOM*|autolearn) : ;;
+
+ # These have somewhat randomly been added and removed, resulting in useless alerts, so ignore them.
+ RCVD_IN_DNSWL_MED|DKIMWL_WL_HIGH) : ;;
+
SPF_HELO_NEUTRAL)
# some of my domains use neutral spf, treat them the same.
results[SPF_HELO_PASS]=t
keys=(DKIM_SIGNED DKIM_VALID{,_AU,_EF} SPF_HELO_PASS SPF_PASS TVD_SPACE_RATIO)
if [[ $to == *@gnu.org && $from == *@gnu.org ]]; then
keys=(ALL_TRUSTED TVD_SPACE_RATIO)
- elif [[ $to == *@gnu.org ]]; then
- # eggs has RCVD_IN_DNSWL_MED
- keys+=(RCVD_IN_DNSWL_MED)
- elif [[ $from == *@gnu.org ]]; then
- # eggs has these
- keys+=(RCVD_IN_DNSWL_MED DKIMWL_WL_HIGH)
+ # from eggs had DKIMWL_WL_HIGH sometime in 2022, then DKIMWL_WL_MED unti march 2023
fi
for t in ${keys[@]}; do
fi
done
if (( ${#results[@]} || ${#missing[@]} )); then
- printf "$HOSTNAME spamtest %s/%s\n" "$latest"
+ printf "$HOSTNAME spamtest %s\n" "$latest"
if (( ${#results[@]} )); then
printf "unexpected %s" "${!results[*]} "
fi
if (( ${#missing[@]} )); then
printf "missing %s" "${missing[*]}"
fi
- echo
- echo mailtest-check: cat $latest:
- cat $latest
- echo mailtest-check: end of cat
- printf "$(tput setaf 5 2>/dev/null ||:)█$(tput sgr0 2>/dev/null||:)%.0s" $(eval echo "{1..${COLUMNS:-60}}")
+ echo # ends our printf string buildup
+ cat $resultfile
+ echo mailtest-check: end of spam debug results
+ # lets just handle 1 failure at a time in interactive mode.
+ if $int; then
+ echo mailtest-check: from: $from, to: $to
+ exit 0
+ fi
+
+ # less verbose debug output, commented since I might want it another time.
+ # if $int; then
+ # echo mailtest-check: cat $latest:
+ # cat $latest
+ # echo mailtest-check: end of cat
+ #fi
fi
+ rm -f $resultfile
+ for r in ${results[@]}; do
+ case $r in
+ # iank: for when we want to handle dns errors differently.
+ # also uncomment declaration of dnsfail above.
+ # DKIM_INVALID|T_SPF_TEMPERROR|T_SPF_HELO_TEMPERROR)
+ # dnsfail+=1
+ # ;;
+ *)
+ unexpected=$(( unexpected + 1 ))
+ ;;
+ esac
+ done
+ for miss in ${missing[@]}; do
+ # At some point we had annoying dns failures that we couldn't solve so we
+ # we counted dns fail related results separately and alert differently.
+ # DKIM_VALID|DKIM_VALID_AU|DKIM_VALID_EF|SPF_HELO_PASS|SPF_PASS|
+ case $miss in
+ *)
+ unexpected+=1
+ ;;
+ esac
+ done
+ mapfile -O ${#p_missing_dnswl[@]} -t p_missing_dnswl <<EOF
+mailtest_check_missing_dnswl{folder="$folder",from="$from"} $missing_dnswl
+EOF
+ mapfile -O ${#p_unexpected_spamd_results[@]} -t p_unexpected_spamd_results <<EOF
+mailtest_check_unexpected_spamd_results{folder="$folder",from="$from"} $unexpected
+EOF
fi # if spamdpid
fi # if $slow
- fi # if [[ $latest ]]
-
- now=$(date +%s)
- limit=$(( now - 60 * min_limit ))
- age_sec=$(( now - last_sec ))
- e $((age_sec / 60)):$(( age_sec % 60 )) ago. to:$to from:$from $latest
-
- if (( last_sec <= limit )); then
- echo $HOSTNAME mailtest $folder $from $(date -d @$last_sec +'%a %m-%d %H:%M')
- fi
- # usec = unix seconds
- pr <<EOF
+
+ now=$EPOCHSECONDS
+ age_sec=$(( now - last_sec ))
+ e $((age_sec / 60)):$(( age_sec % 60 )) ago. to:$to from:$from $latest
+
+ # usec = unix seconds
+ mapfile -O ${#p_last_usec[@]} -t p_last_usec <<EOF
mailtest_check_last_usec{folder="$folder",from="$from"} $last_sec
EOF
+ done # end for from in ${froms[@]}
+ done # end for folder in ${folders[@]}
+
+ dir=/var/lib/prometheus/node-exporter
+ path=$dir/mailtest-check.prom.$$
+ if $doprom && [[ -e $dir ]]; then
+ for l in "${p_unexpected_spamd_results[@]}"; do
+ printf "%s\n" "$l" >>$path
+ done
+ for l in "${p_missing_dnswl[@]}"; do
+ printf "%s\n" "$l" >>$path
+ done
+ for l in "${p_last_usec[@]}"; do
+ printf "%s\n" "$l" >>$path
+ done
+ mv $path $dir/mailtest-check.prom
+ # note: node_textfile_mtime_seconds will tell us when this last happened. useful for debugging.
+ fi
+}
+
+loop-main() {
+ # When running under systemd, the system just started. Ve nice and
+ # give programs some time to finish their startup.
+ sleep 10
+ while true; do
+ premain_sec=$EPOCHSECONDS
+ main
+ sleep $(( 300 - ( EPOCHSECONDS - premain_sec ) ))
done
-done
-if $slow; then
- pr <<EOF
-mailtest_check_unexpected_spamd_results $unexpected
-EOF
+}
+
+
+if [[ $INVOCATION_ID ]]; then
+ loop-main
+else
+ main
fi
-mv /var/lib/prometheus/node-exporter/mailtest-check.prom.$$ /var/lib/prometheus/node-exporter/mailtest-check.prom
+
+exit 0