The online score is dropped when your node has offline audits in the checking window, i.e. when the satellite is come to audit your node it was not online - didn’t answer on audit request.
So, you should figure out why it could happens. I have a few thoughts:
your router dropping packets from time to time (could be some anti-spoofing protection enabled on your router)
I don’t have a single error in the log. I started the log anew two days ago to narrow it down, but nothing can be found, not even at the time the reconnect takes place.
I think I can rule out a fault on the router because this problem takes place in two different locations, i.e. two different networks. DDNS only runs once directly in the FritzBox with myFritz DDNS.
If the node were overloaded, errors should appear in the log, right? But I can’t imagine that since all nodes run their own Pi and the system load is not really high.
My monitors also show no down time. If something were to be blocked, would they have to recognize it or not? And I monitor my nodes with 5 different tools. Uptimerobot, Freshping, Hetrixtools, Zabbix and Netdata
@Alexey
wouldn’t that be covered if he ran a successrate.sh on the log and got 100% audits…?
@LinuxNet
if so, which i believe is the case running successrate.sh would be much easier and more extensive, ofc you may be highly skilled in AWK or searching on multiple terms conditional term or whatever it would be called.
but i know i would be pretty lost trying to do such a search, tho if i had to i would do some random sampling using audit search to find them… i’m not really good friends with grep yet…
trying a different DDNS is a pretty good idea, that would be a difficult problem to spot from a local perspective…
you could also just run a continuous ping on your own ddns address see if it’s always responding… ofc networks being rather smart it might not be 100% sure it’s accurate as you local dns might do some sort of caching and just basically end up testing your internet connection.
nope. It’s just check that there are downloaded GET_AUDIT, that is. However, you can count all GET_AUDIT which is “started” and substract failed and downloaded GET_AUDIT, if result not zero, then you have started audits which is never finished.
@Alexey Is there a special command with which I can search or what I can search for?
My online time continues to decrease
But I have to withdraw my statement that I never received an offline report. I looked through all of my monitoring tools and found that the node is offline relatively often because of the reconnect. Sometimes it’s 2 minutes, sometimes 10.
I tested it and it actually takes a long time until I can reach the services behind my DDNS again. If I take the new IP, of course, it works immediately.
I will test another DDNS provider right away, I was at my second home for the last 2 days where my second node is. This also runs with a dynamic IP and MyFritz. And here, too, there are relatively frequent reports that the node is always offline for some time after a reconnect.
But what do I do if this continues and another DDNS does not bring any improvement? Then I can’t do anything about it and eventually reach the critical mark … or is there something like a “reset” at the end of the month?
why would your ip address change so often to begin with… i mean sure a dhcp setup usually lease an ip address out for like a week and only changes if the device is offline when the ip lease runs out i think…
else the dhcp type ip leases are very long term… sure i guess it would be very different for global ip’s as an isp could maybe setup some sort of configuration where it would cycle through ip’s to optimize their usage, maybe…
the ddns thing i suppose would be a question of latency, i don’t think dns tables are live updated, which is something that is painfully seen in web domain names and such, usually goes so slow.
2-10 minutes to update a ddns table doesn’t seem unreasonable… you have to remember this needs to be accessible over the internet on the entire globe…
i think the answer to solving your problem here isn’t to try to improve the 2-10minutes ddns update latency, but looking at why do your ip address cycle all the time… for a server even when running on a dynamic ip type solution, it shouldn’t really change, aside from if you got stuff that keeps rebooting… like say vpn that reconnects often, unstable internet… [insert many different options here which would lead to unstable connections]
i duno how slow acting ddns services are tho… but it doesn’t seem slow from what i know about this stuff.
I can’t change that myself unless I book a business line. I can only select a time window in my FritzBox when this reconnect should take place. I can’t turn it off and every DSL connection here has it. With cable it’s different. They have a fixed IP for a relatively long time that rarely changes.
I hope I understood you correctly.
My internet is absolutely stable.
with a ddns provider configured in your fritzbox, the ip should be updated within a few minutes because the fritzbox sees the new ip and pushes an update. DDNS providers usually have a TTL of 60s on the dns entries so within a minute the new ip should be seen by all clients too.
Won’t get better than that.
I suppose the problem is solved… that uptime has a little flux i don’t think is really an issue… ofc since it’s a new feature it might be a bit over sensitive…
keep an eye on it and try to track the latency of the switch long term… i mean we are allowed what 0.7% downtime at present which is 5 hours a month, so if we say 10min pr day, so 6 days in an hour and so 30 days… so borderline… but you should be fine especially since its unlikely to be 10 minutes…
ofc it doesn’t allow much room for error, but not sure how strictly it’s enforced atm…
i think you can be offline for like 12 days straight before you get a suspension…
ofc thats for now… but who says you won’t find a way to fix it before stricter restrictions comes into play… and really i don’t think the 0.7% downtime allowed per month will change…
5 hours pr month seems fairly sensible.
i wouldn’t worry about it,
the worst that can happen is that you get a suspension i think…
5 hours downtime is an old value and doesn’t mean anything at the moment. The final value will be a lot higher.
Furthermore there are no consequences for downtime at the moment. The system is in the testing phase.
i checked it in the documentation, and even tho it’s not active doesn’t mean it’s not relevant to linuxnet’s node future, since even with his issues it should keep him within the allowed tolerances.
when they become relevant… thus making any effort towards fixing his problem, kinda more a state of pride rather than need.
I will definitely continue to watch this and I am curious what happens at the turn of the month.
I also wrote an email to my DSL provider and asked whether there was a possibility of getting a fixed IP apart from the business tariff or at least minimizing the forced disconnection and no longer having it daily.
I also don’t think that another DDNS can help because, in my experience in recent years, the FritzBox’s MyFritz service is the fastest there, as @kevink says, the FritzBox immediately pushes the new IP. I tested it again yesterday, restarted a Fritzbox, then had a new IP and the domain immediately resolved the new IP. So it always seems to be different, I’ve had the case a few times that the IP was not updated so quickly.
Nonetheless, I of course entered a different DDNS on one of two FritzBoxes for testing. I will give you feedback again. Thank you all.
you might be able to lobotomize the router… ofc your isp might not like that… but essentially it’s just most likely some sort of BSD based router OS, so really its just a matter of how you get it to not reboot… because i bet if the router stopped rebooting, it would stop getting a new ip… ofc it might also stop updates and cause other issues…
