Where are the requests coming from? ip?
My result 350-850ms (once 3500!!!)
I’ve got weird tests result for this node:
https://supernodes.nem.io/details/947
i passed 3098 round but then i failed 2 rounds and the reason is Responsiveness.
You mentioned that anti-dos adjustments can affect this test. For instance, i used
fail2ban
and
sudo sed -i -e ‘$i et.ipv4.tcpsynretries = 3’ /etc/sysctl.conf
sudo sed -i -e ‘$i et.ipv4.tcpmaxorphans = 16384’ /etc/sysctl.conf
sudo sed -i -e ‘$i et.ipv4.tcpmaxtw_buckets = 16384’ /etc/sysctl.conf
sudo sed -i -e ‘$i et.ipv4.tcp_retries2 = 10’ /etc/sysctl.conf
sudo sed -i -e ‘$i et.ipv4.tcptwreuse = 1’ /etc/sysctl.conf
sudo sed -i -e ‘$i et.ipv4.iplocalport_range = 16384 65535’ /etc/sysctl.conf
sudo sysctl -p
Can these settings prevent my node to pass the test?
Can i somehow initialize those test to see if they will pass or no? Because when we can see the results once in 6 hours and after each restart we need to wait 2-3 rounds to get a proper results it demand a lot of time and nervous to reach the final result.
The responsiveness test issues 10 concurrent height requests to the node in question. The requests are done by the master which is located in the US. I don’t know if fail2ban can intefere with that test. i never tried.
https://supernodes.nem.io/details/947
That looks like NIS is not running any more. How much RAM are you giving to NIS?
can this test be initiated by me or by you or with some api request when needed to check the node ? because it’s crazy to wait 12 hours or more each time to check a guess…
this node runs on vultr vps located in usa as you can see ping and bandwidth are both ok.
the node is running: http://45.77.185.3:7890/node/info
but indeed it crashed before. the reason was lack of memory.
server has 4gb ram and before crash 3gb were allocated to the nis.
after the crush i setup swap file and reduce memory allocation for the nis to 2.5gb.
@uox: The failure was most likely due to the NIS crash. If you want to simulate you would have to write some script that can initiate 10 height requests concurrently. Not sure how much skills you have wrt coding.
What was your node’s alias again? ^^
Sberbank-node-1
There is only ‘sberbank-node-3’ and ‘sberbank_gate2’.
so failure in last round appeared just because i restarted the nis and now i need to wait another 2 rounds to check if it works or no?
Sorry.
Sberbank-node-3
nis.transactionHashRetentionTime = -1
Can it affect?
well i’m not a programmer but the problem has to be solved … so can you give me more details on what exact requests that script has to send to node ?
That node seems to be not so well connected to the US. But it should pass most of the time. Occational failure might be due to the connection from US to Russia being not so good.
The config setting have no influence on the test.
yes, because NIS was down. If you can keep NIS up it will not fail, it usually has 100ms for the responsiveness test which mean it has a good connection to the master.
roger you ! will wait for another round then. thank you !
can you please give me any advise about memory allocation for nis on the server with 4gb in total ?
2500M should be ok, that is what i use on some servers.
Does hosting a supernode enable faster TPS to the immediately connected contributors? i.e. if 3,000 contributors hold 1,000 XEM’s does that mean they can have a higher TPS between these users? @Saul, @GodTanu, @nzminer would love to know your thoughts .
Cheers,
Alex
No, it has nothing to do with the TPS that the network can do. The max TPS is limited by the block size and is currently set to 2 txes / sec (i.e. 120 txes per block).