EVE-NG performance unusable on Cisco UCS blade

Before posting something, READ the changelog, WATCH the videos, howto and provide following:
Your install is: Bare metal, ESXi, what CPU model, RAM, HD, what EVE version you have, output of the uname -a and any other info that might help us faster.

Moderator: mike

Post Reply
Posts: 1
Joined: Thu Mar 14, 2019 7:21 pm

EVE-NG performance unusable on Cisco UCS blade

Post by enetz » Tue Apr 09, 2019 5:22 pm

I have EVE-NG (2.0.5-1-PRO) running in a VM with 32 vCPU's and 128GB of RAM running on a dedicated Cisco UCS B230-M2 blade with two Xeon E7-2860 @ 2.27Ghz (20 cores).

My initial lab only consists of 6 x NX-OSv 9k nodes and 4 x CSR 1000V (Denali and Everest) nodes. 10 nodes total. Nodes take over 5 to 10 mins to boot. Nodes are slow and unresponsive on the CLI. Nodes crash constantly. The load average in Linux is over 25.

This thread https://eve-ng.net/forum/viewtopic.php?f=3&t=992&p=3741 suggests to change machine type to q35, I don't know what that means, but the change no longer boots the node.

Are there any suggestions to make EVE-NG usable? I need to run at least lab 20 nodes but it can't even handle 10.

Uldis (UD)
Posts: 2276
Joined: Wed Mar 15, 2017 4:44 pm
Location: London

Re: EVE-NG performance unusable on Cisco UCS blade

Post by Uldis (UD) » Wed Apr 10, 2019 7:50 pm

It is not EVE issue and not your box..
Just DO NOT start it at once !
KVM virtualization does not act like vm ware. KVM virtualization share all CPUS at the start process, thats why all become mess and slow.
Your box is capable to run it for sure, just be calm !!!

start 2-3 nodes and wait till boot it fully, then start next..2-3 nodes...etc

only newest family of CPU can hold mass start for heavy nodes like this.
Even lower CPU run this stuff..

And make sure you have disabled UKSM in status of eve.
EVE GUI main window: Tab: System/system status.
switch disable UKSM

Just be calm


P.S. On my esxi server x24 vCPU and 80Gb RAM, it does and run such stuff, simply no rush

Post Reply