0

I'm in the process of migrating to mongo 3.2 to use wiredtiger. previously, I backed up to s3 by looking at dbfile names in the directory and just uploading what I don't have already on s3.

That's not an option with wiredtiger. the file names don't really clearly relate to my db names. What is the best way of incrementally backing up the wiredtiger db?

dae park
  • 26
  • 2
  • Have you ever tried to do a restore with that method? – Markus W Mahlberg Mar 01 '16 at 21:25
  • I am bit confused as coping files is a full backup, not incremental Incremental means a delta from last backup to today's date. So - just to clarify are you interested in full backups (have all in one place) or a kind of incremental one? – profesor79 Mar 01 '16 at 21:42
  • So previously I was creating new dbs every hour or so because I had 5 suffixed on the db name. with previous storage engine db files were created with the same db name in the directory, and all I had to was upload that file when the next suffix was created. But with WIredTiger, the files look different, and I can't really map them to the dbs. – dae park Mar 08 '16 at 20:43
  • To add more details, we did it this way so we can bring 1 particular db to the local ( 1hr worth of db is still like 4gb) for debugging purposes. (also ease of back up and restore as it's all sorted by chronological order.) doing directoryperdb option on mongod kind of solves this issue, but there is still a problem with backing up incrementally, and just bringing back 1 particular db. My understanding is that _mdb_catalog is some sort of metadata mapping of collection and indexes to the data files. – dae park Mar 08 '16 at 21:05

0 Answers0