The following opt-in features would add additional check to check for storage layer corruptions of collections.
- Upon write read what data was committed to disk.
- Periodic or scheduled scanning of a collection. Similar to collection.validate but non blocking.
wiredTiger open files usage
Currently WT uses a file per collection and index, leading in some scenarios to extremely high number of open files/dhandles.
Is there any plan to support one file/dhandle per database?4 votes
A potential workaround could be to split the data, store using Decimal128, and recombine using Aggregation Framework. However, this would add performance and programming overhead that encourages customers to select alternatives to MongoDB.1 vote
I'd like to switch my database from Snappy to Zstd compression.
Currently, it doesn't seem that it is possible to change the compression of an existing collection.
It would be nice if there were a way to do this, even if (for example) it required making a new replica set member which "re-compresses" the data to the new algorithm while cloning it.
As per SERVER-67726, the only way to do this today is to create a new collection and manually copy with mongodump/mongorestore. This doesn't seem to be a viable option, for uptime / data consistency reasons.1 vote
- Don't see your idea?