Replies: 2 comments 2 replies
-
|
@Wojtek242 Can you share more details, eg. host system, number of collected metrics, and maybe drill-down a little bit on the CPU usage of netdata vs. its plugins? This would give us a better idea where to focus. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the reply! I am not too sure what exactly would help the most but I'm happy to iterate. I will focus on host 1 for now as I suspect the root cause is likely the same. Having looked at the metrics, the culprit is likely netdata + apps.plugins + go.d.plugin (go.d.plugin being only present in v2). The avg. CPU utilisation in v2 is 7.5% while in v1 it looks like it hovers around 3-4%. Host 1 is a physical server running Debian 12. When I run v1 (1.37.1) I run the Debian repo version. When I run v2 (2.8.5) I run the netdata repo version. Netdata's v2 system info shows On v1: Every second, Netdata collects 3,263 metrics on yggdrasil, presents them in 631 charts and monitors them with 267 alarms. 
On v2: (not sure where to find the same info on number of metrics, but the setup is unchanged, using default config + config I posted in the original post). 
|
Beta Was this translation helpful? Give feedback.
-
|
@Wojtek242 hey! You might want to check |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
I've been using netdata for the past several years. I've just upgraded to v2 and I see that when my machines are idle resource usage nearly doubles. For one machine from ~5 to ~10% (host 1) on another from ~2.5 to ~7% (host 2). This is a pretty significant resource usage jump. I really liked how netdata allowed me to collect so many metrics with so little overhead, but the overhead is just not so low anymore. Are there any suggestions on how to reduce resource consumption in v2? I've already applied the basics from the guides when using v1 so I'm not interested in how to make netdata, in general, less resource hungry. I want to know how to go from v2 to v1 overhead with ideally the same config.
For context my configs are pretty simple.
For host 1:
And for host 2 (it streams to host 1):
Beta Was this translation helpful? Give feedback.
All reactions