[Metrics UI] Enhanced host details - Processes

[sorantis]

Add Properties tab to Enhanced Host Details

Process summary

• system.process.summary.dead
• system.process.summary.idle
• system.process.summary.runnin
• system.process.summary.sleeping
• system.process.summary.stopped
• system.process.summary.total
• system.process.summary.zombie
• system.process.summary.unknown

Process table

State	Command	Time	CPU %	Memory
system.process.state	system.process.cmdline	system.process.start_time	system.process.cpu.total.pct	system.process.memory.rss.pct

Expanded process row

• system.process.cmdline (Full name)
• process.pid
• process.ppid
• process.args
• user.name
• system.process.fd.open
• CPU graph (last 15min)
• Memory graph (last 15min)
• For APM process: View in APM link

APM correlation

host.name, process.pid, @timestamp can be used to correlate a process with an APM service, that can place an APM agent language icon beside the process name on the table.

[elasticmachine]

Pinging @elastic/logs-metrics-ui (Team:logs-metrics-ui)

[sorantis]

@fearful-symmetry could verify that everything's correct?

[alex-fedotyev]

@formgeist, @sqren - I think that in 7.11 it would make sense to link this page to the new service overview with a kuery bar filter to match host name and PID.
What do you think?

[sorenlouv]

I think that in 7.11 it would make sense to link this page to the new service overview with a kuery bar filter to match host name and PID.

I think that makes sense 👍
(I assume we are talking about the service inventory page - the one that shows a list of all services)

[sorantis]

Related to querying APM: #80311 (comment)

[alex-fedotyev]

@sqren - I think that since this integration already has a particular PID on particular host, would it make more sense to land in the individual service overview landing page for that service?
This probably means a separate look up needed to be implemented to confirm that APM is collecting that for that PID ahead of the linking.

The user flow would look like this:

• SRE is troubleshooting a high CPU alert on a particular host.
• SRE identified that a particular Java process is consuming most of the CPU.
• SRE switches to APM to troubleshooting that particular Java application.

[sorenlouv]

This probably means a separate look up needed to be implemented to confirm that APM is collecting that for that PID ahead of the linking.

So given a particular PID, link to a service? We can do that if PIDs are unique across all services. But I'm not sure they are, in which case we could have several services matching the same PID. Perhaps if we combine host + PID it has a higher chance of finding a uniquely matching service?

We could make a link like this, that would redirect to the correct service:

/app/apm/link-to/service?host=my-host&pid=my-pid 

# Will result in a redirect to: /app/apm/my-java-service

[alex-fedotyev]

Agreed.
This would have to be HOST + PID type of filter as well as time information.
In this use case, user is starting on the host and looking at the list of processes looking for highest consumption... so processes are expected to be unique.

[Zacqary]

Need some guidance on how to format the running time column. A lot of processes I'm running into during testing are active for multiple hours, so I had to chop off the milliseconds display for them. But listing them along with entries that contain milliseconds looks confusing:

Should we just drop milliseconds and stick with HH:MM:SS for consistency? Or something else?

[Zacqary]

Are those little lines supposed to be dynamically generated from the data, or are they just graphics?

[sorantis]

1. we can drop the milliseconds
2. these are real data points