// 1. Configure settings below
// 2. Run with "mongo db-spotchecks.js"

uriOld = "mongodb://user:pass@hosname:27017/admin?replicaSet=MySet&ssl=true"
uriNew = "mongodb://localhost"
dbName = "MyDB";
excludeCollections= ["CronLog", "HitLog", "Freq", "SystemLog", "Message"] // Exclude any non-critical collections
sample = {
  min: 10,
  max: 2000
};

// connect to mongos
connOld = new Mongo(uriOld);
dbOld = connOld.getDB(dbName);

connNew = new Mongo(uriNew);
dbNew = connNew.getDB(dbName);

var collPattern = new RegExp("^(" + excludeCollections.join("|") + "fs|system)\.");
print("regexp=" + collPattern.toString());

// For each collection, pick 10 docs at random and compare
dbOld.getCollectionNames().forEach(function(collname) {
  if (!collPattern.exec(collname)) {

    // Check count and size first
    cstatsOld = dbOld.runCommand( { collStats : collname  } )
    cstatsNew = dbNew.runCommand( { collStats : collname  } )
    // printjson(cstatsOld);
    // printjson(cstatsNew);
    if (cstatsOld.count === cstatsNew.count) {
       print("Count " + cstatsOld.count + " OK");
    } else {
       print("ERR Count mismatch " + cstatsOld.count  + " -> " + cstatsNew.count) ;
    }
    if (cstatsOld.size === cstatsNew.size) {
       print("Size " + cstatsOld.size + " OK");
    } else {
      print("ERR Size mismatch " + cstatsOld.size  + " -> " + cstatsNew.size) ;
    }

    fails = 0;
    // sample between 10 and 2000 docs
    count = dbOld[collname].count();
    sampleSize = Math.floor(Math.min(sample.max, Math.max(count * 0.001, sample.min)));
      print(collname + " count="+ count +" with sampleSize="+sampleSize);
    dbOld[collname].aggregate(
      [{
        $sample: {
          size: sampleSize
        }
      }]
    ).forEach(function(docOld) {
      docNew = dbNew[collname].findOne({
        _id: docOld._id
      });
      stringOld = JSON.stringify(docOld);
      stringNew = JSON.stringify(docNew);
      if (stringOld == stringNew) {
        // print("Doc matches: " + docOld._id);
      } else {
        print("Doc mismatch: " + docOld._id);
        print("Old=");
        printjson(docOld);
        print("New=");
        printjson(docNew);
        fails++;
      }
    });
    if (fails > 0) {
      print(fails + " fails");
    } else {
      print("All OK");
    }
  }
});