Few changes to the metricbeat exported fields #2101
Conversation
monicasarbu
force-pushed
the
few_changes_to_the_metricbeat_exported_fields
branch
from
fd05079
to
5a7d9d9
Compare
monicasarbu
added
review
and removed
in progress
| "5": system.Round(loadStat.Load5, .5, 4), | ||
| "15": system.Round(loadStat.Load15, .5, 4), | ||
| }, | ||
| "norm": common.MapStr{ |
There was a problem hiding this comment.
I had to look up what the difference is between avg and norm here. As far as I know, the periods are in minutes. So it could be for avg: system.load.minute(s).1. Not sure how per core would fit into this model.
Alternatively what about system.load.1 and system.load.core.1 ?
There was a problem hiding this comment.
Thinking about it again, norm is potentially better then I initially thought. it means, all the values here will be between 0-1 ? Third option: system.load.1 and system.load.norm.1. I'm curious if we should ship the number of cores as part of load, so the calculations to "normalise" could happen on the kibana side.
There was a problem hiding this comment.
That's a good idea, to keep system.load.1. I will do the changes.
|
@monicasarbu For |
| @@ -21,6 +21,9 @@ https://github.com/elastic/beats/compare/v5.0.0-alpha4...master[Check the HEAD d | |||
| - The undocumented file output `index` setting was removed. Use `filename` instead. {pull}2077[2077] | |||
|
|
|||
| *Metricbeat* | |||
| - Create a separate metricSet for load under the system module {pull}2101[2101] | |||
There was a problem hiding this comment.
It should also be mentioned, that load was removed from CPU.
monicasarbu
force-pushed
the
few_changes_to_the_metricbeat_exported_fields
branch
2 times, most recently
from
0f8064a
to
c32fe54
Compare
| { |
There was a problem hiding this comment.
The previous example data had better example values. The new ones are all 0.
|
LGTM. I just left a minor comment that can be addressed at another time if you like. |
- system.memory.actual.free is replaced with system.memory.available - system.memory.actual.used is not needed as it can calculated from system.memory.total - [ ] system.memory.available - system.filesystem.avail is replaced with system.filesystem.available - add system.load.norm.*
monicasarbu
force-pushed
the
few_changes_to_the_metricbeat_exported_fields
branch
from
638b950
to
b47ab14
Compare
The PR includes the following changes:
loadmetricset in the system modulesystem.load.norm.*system.filesystem.availis replaced withsystem.filesystem.availableNOTE: This PR also replaces
system.memory.actual.freewithsystem.memory.availableandremoves
system.memory.actual.usedas it can be calculated fromsystem.memory.total-system.memory.available. But these changes are reverted in the #2124.