Monitoring with SaltStack VS Zabbix
-
@scottalanmiller said in Monitoring with SaltStack VS Zabbix:
At the moment, you'd still want something like Zabbix for monitoring if you want real time alerts. Sodium will be getting there, but isn't there yet. So you'll need to retain that old infrastructure for the time being, at least for things like servers.
Sounding like a pre-marketing spin-up?
-
@Obsolesce said in Monitoring with SaltStack VS Zabbix:
@scottalanmiller said in Monitoring with SaltStack VS Zabbix:
At the moment, you'd still want something like Zabbix for monitoring if you want real time alerts. Sodium will be getting there, but isn't there yet. So you'll need to retain that old infrastructure for the time being, at least for things like servers.
Sounding like a pre-marketing spin-up?
No, this is full on marketing.
-
@dyasny said in Monitoring with SaltStack VS Zabbix:
I do all my monitoring with prometheus and grafana. Stats are saved, everything looks nice, and alerts are fully customizable. Can't complain
You should do a write up!
-
@wirestyle22 plenty of those on 'dem interwebz
-
@scottalanmiller said in Monitoring with SaltStack VS Zabbix:
@flaxking said in Monitoring with SaltStack VS Zabbix:
Collecting metrics and triggering alerts based on predefined characteristics would be a whole area of monitoring you would miss out on.
Collecting metrics is a core part of Salt. The events framework is part of Sodium. Those two pieces are there natively when you use SodiumSuite, it's just not extensive (yet).
It's good at exporting values, but you would need to integrate another tool to store the data, analyze it and send alerts based on the data over time. Which I guess is where Sodium could come in.
-
@flaxking said in Monitoring with SaltStack VS Zabbix:
@scottalanmiller said in Monitoring with SaltStack VS Zabbix:
@flaxking said in Monitoring with SaltStack VS Zabbix:
Collecting metrics and triggering alerts based on predefined characteristics would be a whole area of monitoring you would miss out on.
Collecting metrics is a core part of Salt. The events framework is part of Sodium. Those two pieces are there natively when you use SodiumSuite, it's just not extensive (yet).
It's good at exporting values, but you would need to integrate another tool to store the data, analyze it and send alerts based on the data over time. Which I guess is where Sodium could come in.
Exactly. SodiumSuite has been doing that part from the beginning. You still don't want to use Salt as the basis for anything intensive, though. It's meant to be run like 1-4 times an hour, not several times a minute.
-
@dyasny said in Monitoring with SaltStack VS Zabbix:
I do all my monitoring with prometheus and grafana. Stats are saved, everything looks nice, and alerts are fully customizable. Can't complain
What about for monitoring things that aren't naturally represented as metrics?
-
@flaxking give me an example
-
@flaxking said in Monitoring with SaltStack VS Zabbix:
@dyasny said in Monitoring with SaltStack VS Zabbix:
I do all my monitoring with prometheus and grafana. Stats are saved, everything looks nice, and alerts are fully customizable. Can't complain
What about for monitoring things that aren't naturally represented as metrics?
What kind of monitoring?
-
@Obsolesce said in Monitoring with SaltStack VS Zabbix:
@flaxking said in Monitoring with SaltStack VS Zabbix:
@dyasny said in Monitoring with SaltStack VS Zabbix:
I do all my monitoring with prometheus and grafana. Stats are saved, everything looks nice, and alerts are fully customizable. Can't complain
What about for monitoring things that aren't naturally represented as metrics?
What kind of monitoring?
@dyasny Hmm, well I guess what I was thinking about probably falls under "reporting" rather than monitoring.
-
@flaxking You mean actual graphs and reports on stats gathered over time? Grafana is all about that
-
@dyasny said in Monitoring with SaltStack VS Zabbix:
@flaxking You mean actual graphs and reports on stats gathered over time? Grafana is all about that
Grafana is only the display tool, you need something to fill in those graphs.
-
@dyasny said in Monitoring with SaltStack VS Zabbix:
@flaxking You mean actual graphs and reports on stats gathered over time? Grafana is all about that
I mean like if you wanted to report on all the different OSes you have running - which I didn't think was possible using Prometheus + Grafana without determining a numeric value for each OS, but it looks like I'm wrong about that.
-
I am currently using PRTG for monitoring and alerting. It works very well. I especially like the push alerts to the Android app, as they have saved me countless times over the past 6 years. How well does the alerting work in Zabbix and also in Salt and sodium?
-
@flaxking said in Monitoring with SaltStack VS Zabbix:
@dyasny said in Monitoring with SaltStack VS Zabbix:
@flaxking You mean actual graphs and reports on stats gathered over time? Grafana is all about that
I mean like if you wanted to report on all the different OSes you have running - which I didn't think was possible using Prometheus + Grafana without determining a numeric value for each OS, but it looks like I'm wrong about that.
The only way I can think of doing that is having each OS type in a different job. At least it used to be true that Prometheus doesn't support string metrics. I hadn't seen anything where they've changed that.
-
@flaxking yup, all that is easily possible
-
@stacksofplates said in Monitoring with SaltStack VS Zabbix:
@flaxking said in Monitoring with SaltStack VS Zabbix:
@dyasny said in Monitoring with SaltStack VS Zabbix:
@flaxking You mean actual graphs and reports on stats gathered over time? Grafana is all about that
I mean like if you wanted to report on all the different OSes you have running - which I didn't think was possible using Prometheus + Grafana without determining a numeric value for each OS, but it looks like I'm wrong about that.
The only way I can think of doing that is having each OS type in a different job. At least it used to be true that Prometheus doesn't support string metrics. I hadn't seen anything where they've changed that.
That's how we did it.
