Conversation
idegtiarenko
added
>non-issue
:Distributed/Allocation
|
Documentation preview: |
|
Pinging @elastic/es-distributed (Team:Distributed) |
| return builder.startObject() | ||
| .field("shard_count", shards) | ||
| .field("forecasted_write_load", forecastedWriteLoad) | ||
| .humanReadableField("forecasted_disk_usage_bytes", "forecasted_disk_usage", ByteSizeValue.ofBytes(forecastedShardSize)) |
There was a problem hiding this comment.
I think it would be beneficial to also add actual disk usage here.
I think there is a confusion between forecasted disk usage and actual disk usage, may be adding both here would help to highlight the difference
| - is_true: 'cluster_balance_stats.nodes.test-cluster-0' | ||
| - gte: { 'cluster_balance_stats.nodes.test-cluster-0.shard_count' : 0 } | ||
| - gte: { 'cluster_balance_stats.nodes.test-cluster-0.forecasted_write_load': 0.0 } | ||
| - gte: { 'cluster_balance_stats.nodes.test-cluster-0.forecasted_disk_usage_bytes' : 0 } |
There was a problem hiding this comment.
I changed the structure (tiers are now nested under tiers field and added a new nodes field in the cluster_balance_stats) as well as renamed (total_write_load -> forecasted_write_load and total_shard_size -> forecasted_disk_usage). I think this better represents the actual values.
The endpoint is internal and has not been released in any version yet, so I think it is okay to do so.
DaveCTurner
left a comment
DaveCTurner
left a comment
There was a problem hiding this comment.
Looks good, one (global) naming nit is that I'd prefer forecast over forecasted throughout. See e.g. https://www.techtarget.com/whatis/feature/Forecast-or-forecasted.
3 participants
This PR adds a per node shard count, forecasted ingest load and forecasted shard sizes to the internal stats endpoint.