As I said before: if you can show real examples of the complete metrics,
with their full sets of labels, then we may be able to help.
What do I mean by this? For example, if I go into the PromQL web interface
and type "node_filesystem_avail_bytes" as the query, I get the following:
Could you provide some help on this topic?
On Tuesday, December 7, 2021 at 10:17:59 AM UTC+8 nina guo wrote:
> in container_cpu_usage_seconds_total, there are following labels:
> instance
> name
> kubernetes_io_arch
> hostname
> job
> cpu (the value of this label is "total")
>
>
> On Tuesday,
in container_cpu_usage_seconds_total, there are following labels:
instance
name
kubernetes_io_arch
hostname
job
cpu (the value of this label is "total")
On Tuesday, December 7, 2021 at 10:15:15 AM UTC+8 nina guo wrote:
> Thank you.
>
> in container_cpu_usage_seconds_total, there are following
Thank you.
in container_cpu_usage_seconds_total, there are following labels:
instance
name
kubernetes_io_arch
hostname
job
On Monday, December 6, 2021 at 5:54:08 PM UTC+8 Stuart Clark wrote:
> On 06/12/2021 08:43, nina guo wrote:
> > May I ask how to check all the available labels for this?
On 06/12/2021 08:43, nina guo wrote:
May I ask how to check all the available labels for this? Can we
define new labels?
For instances, we use file discovery to get the monitored target list.
Just enter the metric name in the query page of the Prometheus UI and
you'll see all the different
May I ask how to check all the available labels for this? Can we define new
labels?
For instances, we use file discovery to get the monitored target list.
On Monday, December 6, 2021 at 4:29:27 PM UTC+8 Stuart Clark wrote:
> On 06/12/2021 08:19, nina guo wrote:
> > Thank you for your reply.
>
On 06/12/2021 08:19, nina guo wrote:
Thank you for your reply.
Let me clarify what we are going to do more.
We have the following alert rules for a container. But we still want
to include the CPU usage of the corresponding host which host this
container. That is:
- when CPU usage for
Thank you for your reply.
Let me clarify what we are going to do more.
We have the following alert rules for a container. But we still want to
include the CPU usage of the corresponding host which host this container.
That is:
- when CPU usage for container is larger than 85%, and CPU usage
Yes, but you need to be really clear on what you're doing. What you need
to realise is that this is a *vector* expression:
expr: containerCPUusage > 80
This may return zero, one, or many results in a vector: that is, a set of
all timeseries with metric name "containerCPUusage" that meet
9 matches
Mail list logo
