CODEWITHSUNDEEP
mongodb
Ouroboros
Ouroboros

Reputation: 1524

MongoDB does not reclaim space despite running repairDatabase()

We have been running mongoDB in single unsharded instance, just one database. The size of data files was 0.45 GB. When I looked into the storageSize of all the collections, the total size was ~85 MB. In a bid to reclaim unused space, we ran repairDatabase(), with understanding that file sizes grow from 64 to 128 to 256 and so on till 2 GB. Since the mongo object data we have (85 MB) can be accommodated in 64 + 128 MB files, we were expecting the 256 MB file to be reclaimed. However, to our surprise, no space was reclaimed.

Can someone let us know the logic based on which we can find how much space would be reclaimed? Essentially, given total disk space a database takes, and given total mongo object data size, can one estimate accurately how much space would be reclaimed?

The following is the db.stats() output as requested in a comment:

> db.stats()
{
        "db" : "analytics_data_1",
        "collections" : 12,
        "objects" : 207223,
        "avgObjSize" : 353.6659347659285,
        "dataSize" : 73287716,
        "storageSize" : 84250624,
        "numExtents" : 43,
        "indexes" : 26,
        "indexSize" : 21560112,
        "fileSize" : 469762048,
        "nsSizeMB" : 16,
        "dataFileVersion" : {
                "major" : 4,
                "minor" : 5
        },
        "ok" : 1
}
>

Upvotes: 1

Views: 319

Answers (1)

Asya Kamsky
Asya Kamsky

Reputation: 42352

The storage FAQ explains that an extra file is always pre-allocated and as soon as you start writing to it, mongod will preallocate the next file.

Repair won't reclaim any space that would normally exist - it can only help if you've deleted a lot of data or dropped some collections.

Disabling preallocation can save you space but will cost you in performance as the file will be allocated when it's actually needed to write to - and that will slow down inserts.

Upvotes: 2

Related Questions