Allow to control recency of ids in conflicts #529
Conversation
With this commit we add a new parameter to the `bulk` parameter source called `recency` that allows to control the bias towards more recent ids when generating id conflicts.
danielmitterdorfer
added
enhancement
esrally/track/params.py
Outdated
| # A recency > 0 biases id selection towards more recent ids. The recency parameter decides | ||
| # by how much we bias. See docs for the resulting curve. | ||
| # | ||
| # idx_range is in the interval [0;1]. |
There was a problem hiding this comment.
I guess there's a typo here, [0;1] -> [0,1]
There was a problem hiding this comment.
It was actually intentional. The German decimal separator is a comma so it could be misinterpreted by a German as 1/10 instead of the interval from zero to 1 (inclusive). However, we can change it as well.
There was a problem hiding this comment.
I changed it anyway now as I've seen that we use the comma notation in all other places.
esrally/track/params.py
Outdated
| idx_range = min(self.randexp(GenerateActionMetaData.RECENCY_SLOPE * self.recency), 1) | ||
| # the resulting index is in the range [0, self.id_up_to). Note that a smaller idx_range | ||
| # biases towards more recently used ids (higher indexes). | ||
| idx = int(round((self.id_up_to - 1) * (1 - idx_range))) |
There was a problem hiding this comment.
Does round() anyway return an int, if the optional ndigits parameter is not passed?
No harm being explicit here, but wondering about it.
There was a problem hiding this comment.
Oh, you are right. I was probably assuming that it would behave similarly to Math#roundin Java, hence the explicit conversion.
With this commit we add a new parameter to the
bulkparameter sourcecalled
recencythat allows to control the bias towards more recentids when generating id conflicts.