X-Git-Url: https://iankelling.org/git/?a=blobdiff_plain;f=mailtest-check;h=d5d30aa6ae2f275654ed391053bbaa1bafb87657;hb=e8d684282c4b3cdd62859217f02711df5947a5c1;hp=02cdfcbcad94027f720563e87d6509c48f9c19f6;hpb=ae10fa08bb841b99b0df8e827735bef08c05f3ca;p=distro-setup diff --git a/mailtest-check b/mailtest-check index 02cdfcb..d5d30aa 100755 --- a/mailtest-check +++ b/mailtest-check @@ -28,11 +28,6 @@ getspamdpid() { done fi } -pr() { - if $doprom && [[ -e /var/lib/prometheus/node-exporter ]]; then - cat >>/var/lib/prometheus/node-exporter/mailtest-check.prom.$$ - fi -} #### begin arg processing #### @@ -57,10 +52,11 @@ if [[ $1 == nonint ]]; then fi #### end arg processing #### - -if ! $int; then - sleep 60 -fi +# we put this in to avoid dns errors that happen on reboot, +# but I want to debug them. +# if ! $int; then +# sleep 60 +# fi # TODO, get je to deliver the local mailbox: /m/md/INBOX @@ -80,6 +76,7 @@ esac 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) @@ -93,7 +90,32 @@ main() { 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 + ### 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 @@ -122,6 +144,9 @@ EOF 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 @@ -131,13 +156,15 @@ EOF fi # webmail sends them to cur it seems while read -r file; do - if [[ $file -nt $latest ]]; then + 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 to=$(awk '/^Envelope-to: / {print $2}' $latest) - last_sec=$(awk '/^Subject: / {print $4}' $latest) if $slow; then if ! $int; then @@ -148,16 +175,18 @@ EOF if [[ $(readlink /proc/$$/ns/net) != "$(readlink /proc/$spamdpid/ns/net)" ]]; then spamcpre="nsenter -t $spamdpid -n -m" fi - + unset results declare -A results # pyzor fails for our test message, so dont put useless load on their # servers. # example line that sed is parsing: # (-0.1 / 5.0 requ) DKIM_SIGNED=0.1,DKIM_VALID=-0.1,DKIM_VALID_AU=-0.1,SPF_HELO_PASS=-0.001,SPF_PASS=-0.001,TVD_SPACE_RATIO=0.001 autolearn=_AUTOLEARN resultfile=$(mktemp) - $spamcpre sudo -u Debian-exim spamassassin -t --cf='score PYZOR_CHECK 0' <"$latest" &>$resultfile + $spamcpre sudo -u Debian-exim spamassassin -D -t --cf='score PYZOR_CHECK 0' <"$latest" &>$resultfile - raw_results="$(tail $resultfile | grep -A1 -Fx /usr/local/bin/send-test-forward | tail -n1 | sed -r 's/^\([^)]*\) *//;s/=[^, ]*([, ]|$)/ /g')" + # note: on some mail, its 1 line after the send-test-forward, on others its 2 with a blank inbetween. + # I use the sed -n to filter this. + raw_results="$(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 @@ -186,8 +215,8 @@ EOF # 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) + # eggs has this. it used to have DKIMWL_WL_HIGH sometime in 2022 + keys+=(RCVD_IN_DNSWL_MED) fi for t in ${keys[@]}; do @@ -218,6 +247,11 @@ EOF 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 @@ -228,6 +262,37 @@ EOF #fi fi rm -f $resultfile + for r in ${results[@]}; do + case $r in + # iank: for when we want to handle dns errors differently + # DKIM_INVALID|T_SPF_TEMPERROR|T_SPF_HELO_TEMPERROR) + # dnsfail+=1 + # ;; + *) + unexpected=$(( unexpected + 1 )) + ;; + esac + done + for miss in ${missing[@]}; do + # We expect dns failures from time to time, so + # we count them separately and alert differently. + case $miss in + # iank: dns fail + # DKIM_VALID|DKIM_VALID_AU|DKIM_VALID_EF|SPF_HELO_PASS|SPF_PASS| + RCVD_IN_DNSWL_MED|DKIMWL_WL_HIGH) + missing_dnswl+=1 + ;; + *) + unexpected+=1 + ;; + esac + done + mapfile -O ${#p_missing_dnswl[@]} -t p_missing_dnswl <>$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 ) )) + sleep $(( 300 - ( EPOCHSECONDS - premain_sec ) )) done }