I wonder if this is significant because it appears to copy the ntp.conf from jffs to entware repo before entware ntp starts. Also it looks like it is stalling entware ntp from starting until routers is finished. Could this be for when the ntp.conf gets edited?.
Try editing your ntp.conf fileLikewise, so I think we are all good, graphs behaving normally ... mystery solved!
Old out dated guide
Maybe adding an NTP server by IP to ntpmerlin's ntp.conf would allow NTP to begin synchronizing while DNS is in the process of starting. If you nslookup time.cloudflare.com, perhaps it will return a close server.
This wouldn't help. The built in ntp failed to sync, so ntpMerlin and thus entware ntpd were not launched
I used to see weird stuff like this when I had too many things going on in /jffs/scripts/service-event because it blocks the original event from completing. More syslog messages would be good to see. I remember you posted them before, but a fresh look might be good.
The only change was to stop an update file call to S77ntpd going out whenever you launched ntpmerlin's menu
#!/bin/sh
if [ "$1" = "start" ] && [ "$2" = "SkynetStats" ]; then sh /jffs/scripts/firewall debug genstats; fi # Skynet
/jffs/scripts/ntpmerlin service_event "$1" "$2" & # ntpMerlin
/jffs/addons/unbound/unbound_stats.sh generate "$1" "$2" & # Unbound_Stats.sh#!/bin/sh
/jffs/scripts/ntpmerlin startup & # ntpMerlin
cru a root_servers "12 4 * * * curl -o \/opt\/var\/lib\/unbound\/root\.hints https://www.internic.net/domain/named.cache" # unbound_manager#!/bin/sh
swapon /tmp/mnt/network/myswap.swp # Skynet
. /jffs/addons/diversion/mount-entware.div # Added by Diversion
/jffs/addons/unbound/unbound_stats.sh startup # Unbound_Stats.sh#!/bin/sh
/jffs/scripts/ntpmerlin ntpredirect # ntpMerlin#!/bin/sh
# shellcheck disable=SC2034
if [ "$1" = "start" ] || [ "$1" = "restart" ]; then
# Wait for NTP before starting
logger -st "S77ntpd" "Waiting for NTP to sync before starting..."
ntptimer=0
while [ "$(nvram get ntp_ready)" = "0" ] && [ "$ntptimer" -lt "300" ]; do
ntptimer=$((ntptimer+1))
sleep 1
done
if [ "$ntptimer" -ge "300" ]; then
logger -st "S77ntpd" "NTP failed to sync after 5 minutes - please check immediately!"
exit 1
fi
fi
if [ -f "/opt/share/ntpmerlin.d/config" ]; then
SCRIPT_STORAGE_DIR="/opt/share/ntpmerlin.d"
else
SCRIPT_STORAGE_DIR="/jffs/addons/ntpmerlin.d"
fi
ENABLED=yes
PROCS=ntpd
ARGS="-c $SCRIPT_STORAGE_DIR/ntp.conf -g"
PREARGS=""
PRECMD="killall ntp && killall ntpd"
POSTCMD=""
DESC=$PROCS
PATH=/opt/sbin:/opt/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
. /opt/etc/init.d/rc.funcwell for starters, whenever ntpmerlin runs it is no longer looking for updates for ntpd at the start. I think that feature was causing an achilies heel if you get what I mean. A weak spot in the overall design.@SomeWhereOverTheRainBow
I saw there had been a minor update but what changed?
"More stable" in what sense?
.
There's no "NTP failed to sync" in either of your syslog outputs.
What I mean is, the 5 minute failure condition isn't being met. The last reboot, the router has only been up for 2m30, which may not be long enough for NTP to sync
