You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We should add unikraft <resource> graph. This new command, available on all resources would allow graphing fields of that resource over time. You would be able to observe those properties as they changed.
-f metrics.rss/resources.memory and -f derivative(cpu_time) do not exist yet.
There are kinda interesting though... and feel related to unikraft/cli#373. They're almost like "virtual fields"? Not really in the data model, but can be computed from it.
We should also support non-integer data. Some ideas:
bools could be graphed as a square signal
enums could be graphed as a histogram (when we have multiple inputs)
We should add
unikraft <resource> graph. This new command, available on all resources would allow graphing fields of that resource over time. You would be able to observe those properties as they changed.We should probably use https://github.com/NimbleMarkets/ntcharts.
This would generate a line graph which would keep updating as more information arrived of the memory usage of
instance-foo:We could also show this against the total amount of memory available (as a solid bar):
What about showing it as a percentage? We need some processing:
We could even allow plotting two instances against each other, one of
instance-fooand another ofinstance-bartogether:We could show CPU usage by computing a time delta (or some other function):
Note
-f metrics.rss/resources.memoryand-f derivative(cpu_time)do not exist yet.There are kinda interesting though... and feel related to unikraft/cli#373. They're almost like "virtual fields"? Not really in the data model, but can be computed from it.
We should also support non-integer data. Some ideas:
bools could be graphed as a square signalenums could be graphed as a histogram (when we have multiple inputs)