nickva edited a comment on pull request #3364: URL: https://github.com/apache/couchdb/pull/3364#issuecomment-782437514
To help test the PR I created a silly script which runs with a local `./dev/run -n1` cluster. It configures replication jobs and shares for three `_replicator` dbs and then continuously samples the proportion of running jobs in each db: https://gist.github.com/nickva/7e86b3df19537a60372217e0f68b693a So for the default parameters it might output something like: ``` ** run stats num dbs: 3, total jobs: 100 rdb1/_replicator 66 66% rdb2/_replicator 23 23% rdb3/_replicator 11 11% ** run stats num dbs: 3, total jobs: 100 rdb1/_replicator 61 61% rdb2/_replicator 31 31% rdb3/_replicator 8 8% ``` With a configuration where the shares are all even (100) but the dbs have an uneven number of jobs, after 5-10 minutes there should be a roughly even share of running jobs from each db, even though one db has 500 jobs and others have less. ```python INTERVAL = 20000 CONFIG = { "replicator": { "max_jobs": "100", "max_churn": "20", "interval" : str(INTERVAL), }, "replicator.shares": { RDB1 : "100", RDB2 : "100", RDB3 : "100", } } REPS = { RDB1 : 200, RDB2 : 300, RDB3 : 500, } ``` ``` ** run stats num dbs: 3, total jobs: 100 rdb1/_replicator 30 30% rdb2/_replicator 34 34% rdb3/_replicator 36 36% ** run stats num dbs: 3, total jobs: 100 rdb1/_replicator 29 29% rdb2/_replicator 33 33% rdb3/_replicator 38 38% ``` ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org
