Hello.
The system was failing nicely for last 2 days 
therefore I have put the gc logs from nis and servant here:
http://37.59.120.16/nis-gc.tgz
http://37.59.120.16/servant-gc.tgz
I hope this can give some clue about that issue.
The node seems to be pretty slow for some reason. In the NIS GC logs for example
43810.332: [GC (Allocation Failure) 660475K->588426K(932352K), 0.1754826 secs]
43835.343: [GC (Allocation Failure) 707722K->639541K(926720K), 0.3423668 secs]
43864.157: [GC (Allocation Failure) 756277K->692846K(932352K), 1.3958661 secs]
43865.553: [Full GC (Ergonomics) 692846K->532509K(932352K), 9.6290874 secs]
43897.568: [GC (Allocation Failure) 649245K->580629K(932352K), 0.2493152 secs]
43901.744: [GC (Allocation Failure) 697365K->630117K(932352K), 0.7356606 secs]
43905.628: [GC (Allocation Failure) 746853K->679693K(932352K), 0.8376398 secs]
43906.466: [Full GC (Ergonomics) 679693K->540315K(932352K), 15.9025371 secs]
compared to some NIS gc log from a 2GB vps of mine
95835.328: [GC (Allocation Failure) 696690K->641030K(932352K), 0.0463035 secs]
95866.295: [GC (Allocation Failure) 762374K->705422K(923648K), 0.1239461 secs]
95866.419: [Full GC (Ergonomics) 705422K->589304K(923648K), 1.3758332 secs]
95896.474: [GC (Allocation Failure) 706040K->648261K(932352K), 0.0403407 secs]
95921.037: [GC (Allocation Failure) 764997K->692438K(932352K), 0.0656866 secs]
95921.206: [GC (Allocation Failure) 809174K->742433K(932352K), 0.2053124 secs]
95921.411: [Full GC (Ergonomics) 742433K->647527K(932352K), 1.3401443 secs]
Can you run some cpu benchmark or ask the provider why it is so slow?
Hi,
I have upgraded my node hardware
4core cpu with 16GB VPS.
NIS is running with
java -Xms3584M -Xmx7G
Is it reliable to give 140 harvesting slots?
Yea, should be no problem.
One thing: With such an amount of RAM i would use G1 garbage collector. This is done by adding the paramters
-XX:+UseG1GC -XX:MaxGCPauseMillis=200
to the start script. Also i don’t see an advantage to set -Xms3584M, just use -Xms7G
Wow. Compared to your node mine is really slow :-/
Is your server also on OVH?
If so would it be possible for you to do one benchmark on your node and send me the results?
I would like to do the same tests on my node with the same tool that you use.
That way I would have some numbers that I can compare show to the provider
and ask him why is there so big difference in performance when the severs are so similar
in configuration.
I don’t know if the provider will do anything if i just state that the node is slow without showing him any numbers.
Yeah just another thing regarding processors.
The processor on the my vps are AMD if you have Intel processor on your vps that could be the reason
that you are getting better CPU performance.
I have done one test 2 years ago with MySql database.
I had quad code amd server and an intel dual code desktop machine.
I run a select query on big chunk of data with joins and groups on both machines and the
dekstop machine outperformed the server by about 2 to 2.5 times on each query that i have run.
liftoff!
Dear Bloody Rookie.
I sent message to you about
change alias “SHINSHI INTERNATIONAL PATENT OFFICE” to "OPIS"
and changed my supernode’s “config.properties” to new alias name, yesterday.
http://157.7.135.123:7890/node/info
I think it might be works correctly.
but did not changes alias name on https://supernodes.nem.io/details/436
I failed anything?
@freigeist: you can do a little test
apt-get install bc
time echo “scale=2000; a(1)*4” | bc -l
it calculates Pi with 2000 decimal places.
Alice2 (OpenVz): 0m3.864s
Alice5 (KVM): 0m1.566s
Not really a benchmark test but should give an impression of how fast your node is.
Super. Thank you.
I had made the test that you have put here.
I had to change it to:
time echo "scale=2000; 4*a(1)" | bc -l
to make it work otherwise gave me always an error.
the results were as follows:
3.145.....
real 0m7.647s
user 0m7.277s
sys 0m0.367sreally bad numbers. 
I will contact the provider regarding the issue.
Thank you very much for your time. I will let you know as soon the issue is resolved.
thank you and So sorry, I’ve faild to send message, might be.
2089f3b669c095a9ebee462b2c323626b60012c9d4e087cbc2e7a286ce372516
is transaction hash.
you have sent
change alias <shinshi international patent office> to <opis>
You should not have "<"or “>” in the message.
Also the current name is in upper case: SHINSHI INTERNATIONAL PATENT OFFICE
You have to use the name exactly as it is in the message, for example
change alias SHINSHI INTERNATIONAL PATENT OFFICE to OPIS
I see. thank you.
Hello.
I have received a simple answer from the provider
that their VPSs are on shared resources so the results I can get any time are dependent on the load on the host that my VPS is on. They advise me to upgrade or to move on something with dedicated resources like dedicated server or public cloud.
Aren’t yours vps also shared or you use something else to host your nis nodes?!
I mean it all worked fine until last 2 weeks ago they started screwing me with this performance. 
From your experience what would be the best option to do 
There are quite a few reliable providers. Probably the easiest would be to pick vultr since you pay hourly and setup / teardown of a node is almost instantly and they use fast KVM virtualization.
I would probably pick a 4GB node which is $20 / month and give NIS 2.5GB (with G1 garbage collector) and the servant 256MB. Debian 8 is my favorite OS.
$20 / month is a probably not as cheap as your VPS you have now but not getting payouts is even more expensive.
Thank you very much for your help.
I have choose to upgrade my current vps to 4GB and 3 cpu cores for now.
I gave 2GB memory to NIS. will see how it works.
BTW. you think that with the release of Catapult the vps
requirements will be lower or higher?
idk. We will see.
My node name is Gambit_WI and it fails about once a week. Do I need to upgrade my VPS plan? My current plan is the $20 plan here https://www.digitalocean.com/pricing/#droplet
It is hard to tell why it fails, during the last 15 days it failed a single test.
But for $20 you get a stronger VPS from vultr (i think 4GB KVM), so i would consider moving the node to that provider (i would choose a node in the USA).
There will be a database maintenance on 2017-12-17. The master needs to be shut down for a while to clean up the database. There will be a delay in the testing, so don’t be surprised if the subsequent payout is later than you think it should be.
1 Like