Get some of your cpu back when being hammered (Discussion)

[gateway]

Hey guys, I wanted to say that we got hammered yesterday and had 8 cores pegged on a few servers. This was due to attacks and a lot of miners.

I was able to bring back some cpu by removing the graph stats container and the related javascript..

this is more of a discussion on server load help.

[martin-key]

this is because of the way graphics are drawn. I have found the issue and almost fix it by changing the framework.

…

On 13 Nov 2017, at 22:29, Gateway ***@***.***> wrote: Hey guys, I wanted to say that we got hammered yesterday and had 8 cores pegged on a few servers. This was due to attacks and a lot of miners. I was able to bring back some cpu by removing the graph stats container and the related javascript.. this is more of a discussion on server load help. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#5>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADrjrcbUxetKMQB1WVjarFO8cqZf2eozks5s2KaTgaJpZM4QcXUs>.

[gateway]

Are the stats being pulled every tick? Could we just do it every x interval to update the chart.. and in between cache the chart data so its not always pulling? .. I think if switched to another graph package it might be wort doing this.. have you looked at highcharts?

[martin-key]

I have made changes to this, but they are experimental. I would share this with all the people eventually if I decide to continue my contribution on the open source z-nomp after all the attacks from the last 2 days.

…

On 14 Nov 2017, at 21:08, Gateway ***@***.***> wrote: Are the stats being pulled every tick? Could we just do it every x interval to update the chart.. and in between cache the chart data so its not always pulling? .. I think if switched to another graph package it might be wort doing this.. have you looked at highcharts? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADrjrQXXXHcOmiI9c3ExXlR5WUF5Lvy8ks5s2eUsgaJpZM4QcXUs>.

[gateway]

That brings up a bigger topic which I think we should open a discussion thread here on how to deal with attacks. I felt the same and their are a few things people should have in place to help. (fail2ban), their is a znomp filter for this. proper ip tables, cloudflare. On the ports for the miners filter html data POST/Get request.. all this crap ties stuff up in the event loop.

Here is some data from newrelic from one of my servers.. a lot of time spent hammering live stats.. I think that should be cached more and not so live.. thoughts? Put in a variable to only get data every x seconds?

oh I would be up for funding for any help.. I think the BTG peeps using should kick in a bit to help move this project forward.. but most importantly optimizations..

[martin-key]

We use cloudflare and with specific settings for caching. We can open threads for these problems, because there is no universal approach but we can share opinion.

…

On 14 Nov 2017, at 21:24, Gateway ***@***.***> wrote: That brings up a bigger topic which I think we should open a discussion thread here on how to deal with attacks. I felt the same and their are a few things people should have in place to help. (fail2ban), their is a znomp filter for this. proper ip tables, cloudflare. On the ports for the miners filter html data POST/Get request.. all this crap ties stuff up in the event loop. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADrjrZ0kgNbr8C1lw16AOZt_wnB0XdFwks5s2ejegaJpZM4QcXUs>.