mtalarico / sampler

Sampler

DISCLAIMER: this is not an official MongoDB product, rather it is experimental code that should be used at your own risk. As with any software, test thoroughly in a lower environment.

Using statistical sampling based on Cochran's sample size to validate two mongodb cluster's sychronization.

Compares

• estimated document counts
• indexes
• sample of documents based on statistical analysis

Sharp Edges

• currently lists all namespaces and keeps them in memory. Except the sampler to blow up on clusters that have a high number of namespaces
• currently compares indexes by name
• currently runs a top level $or query to get the batch on the opposite cluster -- even if the collection is not partitioned and only _id is used

TODO:

• remove hard-coded strings from code, replace with consts
• better help output
• sharded collection validation support
• optimize for $in instead of $or for collections that just query on _id