Why the server suddenly stops to Selectable pingomatic?


Faced with a strange problem on a server Selectel - sometimes the server has stopped responding to all requests from the network (pings, http, ssh), saved only restart from the panel selectee (kve terminal there is also not worked), judging by the logs during the outage time he continued to work honestly.
In silage anything interesting and unusual was not.
On the server spinning elementary GosNII applications due to systemd services (in the beginning I was afraid that the logs overflow, but their restriction did not save), the ipsec tunnel for vneshnego IPAS and small postgre with a rebbit. Tech support throws up his hands and says that they say nothing was found wrong.

Now the server is gone completely silent, not even save the restarts.

I will appreciate any ideas in what direction to look for the problem, thank you in advance!
March 23rd 20 at 18:50
4 answers
March 23rd 20 at 18:52
You toaster ask about it, and a support selectel
Yes, also in parallel, say everything is OK, don't know anything, problems do not see - Joan_Denes commented on March 23rd 20 at 18:55
@Joan_Denes, then the problem must tyknut: screenshots of ping/traceroute
At the same time you can see that the problem is not from them - Leo.Schultz commented on March 23rd 20 at 18:58
@Leo.Schultz, Tyknut need at the time when there is a problem :) - Dante commented on March 23rd 20 at 19:01
March 23rd 20 at 18:54
Good afternoon. Nothing just happens:
1) Remember at what point in time did you first discover the problem. Now, remember, what work was carried out this time ?
2) Follow the periodicity of the problem. If problems arise with a stable frequency (e.g. once a day at midnight) - it's a clue, look for what is happening in the system at this time (for example the cron starts a process).
3) "cut off excess". It's a long time to do, after paragraph 1,2. If a full service operation now can not speak, then sequentially turn off all of your services until the problem disappears.
So You will be able to identify the problem, and this is half the battle.
1) a week after adding services. Prior to that the binaries just run hands.
2) Vozvraschaysya. Week, day, couple hours, 15 minutes, stopped to recover. Dependence on Matrosov, updating services are not installed.
3) Very good idea, thanks, I will try soon! - Joan_Denes commented on March 23rd 20 at 18:57
March 23rd 20 at 18:56
You can try to see the server status at the time of problems through a KVM console, it should work (if not working kick support).

If just falls off the network, then I would have updated the network card drivers from the manufacturer's website, distributions usually are not the fresh wood.
March 23rd 20 at 18:58
If the server is running and can access it - see the logs and think.
If the server is unavailable - kick the tech support of the hosting provider.

Find more questions by tags SelectelUbuntuSystem administration