Add a new tsdb track to test _tsid hashing #464
Conversation
|
The CI tests are failing because the template and to be more precise the mappings include more than 21 dimension fields (`"time_series_dimension": "true"). This is a feature that Elasticsearch still does not support and it is the whole [urpose for having this dataset. The (contender) Elasticsearch implementation I have here actually supports more than 21 dimensions. I will need this track dataset to evaluate impact of such feature. |
| "val": { | ||
| "type": "integer" | ||
| }, | ||
| "d_ufseibmjnkguynqfyclevvweebkoxqgvspbbngkvtrsegqtwuyel_001": {"type": "keyword", "time_series_dimension": "true"}, |
There was a problem hiding this comment.
I will probbaly replace the "true" value here with a Jinja "statement" which uses "false" as default and that we can override by means of track.params.
There was a problem hiding this comment.
Or maybe I can change the maximum number of allowed dimensions.
|
We can't just change the max number of dimensions because indexing would fail anyway due to too large values for dimensions and for the |
| "corpora": [ | ||
| { | ||
| "name": "tsdb_synth", | ||
| "base-url": "https://storage.googleapis.com/rally-tracks/tsdb_synth/", |
There was a problem hiding this comment.
Could this be https://rally-tracks.elastic.co/tsdb_synth/ ? They are the same place at the moment, but https://rally-tracks.elastic.co/tsdb_synth/ goes through the cdn
There was a problem hiding this comment.
Ok...so going through the CDN was probably the cause of the issue about the file not being there I guess...just a result of the delay in distributing the file across the CDN probably.
Include a new TSDB track and dataset. The plan is to use it to test _tsid hashing and maybe impact of compression on dimension fields.