How can change the bulk size with geonames

(andy_zhou) #1

how can change the bulk size with geonames
when i change about .rally/benchmarks/tracks/default/geonames/operations/default.json
i wan't change "bulk-size": 5000 to 10000,but when i change it
i can't running esrally
error: Your local changes to the following files would be overwritten by checkout:
** geonames/operations/default.json**
Please, commit your changes or stash them before you can switch branches.
[ERROR] Cannot race. Cannot update track data in [/root/.rally/benchmarks/tracks/default].
how can change the default bulk size

(Daniel Mitterdorfer) #2

Hi @zqc0512,

Rally tracks are fully versioned so Rally can support benchmarking any version of Elasticsearch between 1.x and the latest master version:

daniel@io:tracks/default ‹master›$ git branch -r
  origin/HEAD -> origin/master

So you can see that there exist multiple branches to support that. Rally will check which version of Elasticsearch you want to benchmark (it will use what you've provided via the --distribution-version command line parameter) and tries to find the best branch. Some examples:

  • For 2.4.4 it will find that the best match is the branch "2" and check that one out.
  • For 5.5.0 it will use "5"

This brings me now to your problem: I assume you made your changes on the master branch but Rally has detected that it needs to switch branches because e.g. you're benchmarking Elasticsearch 2.4.5. When attempting to switch branches it detects that you have uncommitted changes and refuses to continue. To avoid that you should make your changes on the branches that you're interested in. That means:

  • If you want to benchmark 1.x: make your changes on branch "1"
  • If you want to benchmark 2.x: make your changes on branch "2"

and so on.

I recommend that you fork the official repository and work in your fork. Then you can also commit and push changes as you see fit. For further information please check the docs about custom track repositories.


(Christian Dahlqvist) #3

You can now add additional files in the operations and challenges directories and have these loaded on startup. You could therefore create a new file under operations that contain the indexing operation with a different bulk size and then reference this from a new custom challenge created based on the existing one.

(Daniel Mitterdorfer) #4

@Christian_Dahlqvist raises a good point here to which I agree. This is definitely recommended practice. See the eventdata track for an example how this can be implemented in practice.


(system) #5

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.