In MongoDB we have to decide how we store relations to other entities. A central relation in our data model is the checkout process. I decided to store the checkouts as an array to the score where the checkout refers too. Only one checkout document - always the last in the array - can have the 'open' status, i.e. 'score is checked out'. From this document we can get the user who has checked out this score. For convenience we have also a field 'checkedOutByUser' at the score document, in order to immediately see if a score is checked out and by whom. This is kind of redundant, but if we store this all in one 'save' action during checkout/checkin, there should become no inconsistency from this redundancy, because the 'save' is atomic. Technically this criteria could easily be replaced be fetching the last checkout from the checkouts array and get this information from this. Primary reason currently for me to keep this is that I can quickly browse the score documents manually in MongoDB tools, and see which are checked out and which are not.
At the beginning of the app development I was connecting without database name in the URI (see below). I was curiuos what happens if you provide a wrong database name. Astonishingly Mongoose returned no error when connecting. Instead the code which tried to read from a collection just didn't find anything, but again also no error here. I find this a little bizarre.
With the free plan of Mongodb Atlas a backup/restore must be done using the cmd line tools mongodump/mongorestore.
https://www.mongodb.com/docs/database-tools/
creates a binary backup of a database
mongodump --uri <connect-url>/<database>
- <connect-url> is the Mongodb connection string from the env file
- <database> is the database to backup
At the beginning I didn't know what the database was, because it didn't appear in the connection URI I've used for quite a while or elsewhere in the Mongoose connection code. I found it here:
So far I've had no other database rather than "test". That's why probably the database could be omitted in the URI. It seems that the database is somehow like a namespace. So unless there are collections of the same name in different databases, one seem to be able to omit the database in the URI. At least this is my understanding. But of course specifying the database in the URI is much more clearer, and from now on I used it in my env variable.
mongodump will by default backup all collections (here: users, scores, ...) into a folder "dump/<database>" within the current directory.
I created a shell script mongobackup.sh which runs mongodump and copies the backup to a backuped folder
restore from a binary backup created with mongodump
The first test you probably want to do is to restore the backup on a different location i.e. database (because you don't want to overwrite your data during this test). It became not immediately clear how to achieve this, but I ended up with this procedure:
- provide the backup in a folder dump/<target-database> within your current directory. So e.g. if with mongodump you ended up with a dump/test folder (because "test" is your database name), provide e.g. folder dump/testrestore.
- execute this command
mongorestore --uri <connect-url> --dir dump/\<target-database\>
It's just that easy. The database "testrestore" will be created if it was not existing.
Important: don't leave the "test" folder inside the dump, otherwise mongorestore will restore to your "test" database too.
Installed on pc-ms4, but today connection timeout while connection via Atlas website was working. However, connection worked from Home PC, so it's possible due to some Quanos restriction.
In the Atlas GUI you can only delete one document at a time. In order to delete documents from a filter query, use Compass.