Skip to main content
The documentation in this section is deprecated and refers to a product that was retired on October 15, 2024. LiteFS itself is not affected by the retirement of the LiteFS Cloud service. Learn more about how to disconnect from LiteFS Cloud and other options for disaster recovery in our community post: Sunsetting LiteFS Cloud.
Important: LiteFS is outside our scope of support. We are not able to provide support or guidance for it. Use with caution.

ā€‹
Overview

You can restore your LiteFS database to any point in time in the last 30 days, with 5 minute granularity. This means that when you select a time, and LiteFS Cloud will find the most recent snapshot from before that timestamp, which is guaranteed to have been the state of the database no more than 5 minutes before the timestamp you selected. Note: if you decide that you made a mistake restoring from snapshot, you can also get your pre-restore state back, simply by restoring again, and selecting the time right before you did the first (erroneous) restore!

ā€‹
Restoring your LiteFS database

To restore your LiteFS database to a snapshot, navigate to the Fly.io LiteFS dashboard, find the LiteFS Cloud cluster and database you would like to restore, and click the ā€œview backupsā€ link. Then you can select a date and time to restore from. Since restore is a destructive function (you are likely to lose up to 5 minutes of data if you need to ā€œrevertā€ the restore), youā€™ll need to type the name of the cluster and database to make sure youā€™re restoring the database you intend. The restore normally takes just a few seconds, or up to a couple of minutes if your database is very large. However, you wonā€™t see the new data until after you make a write to your database. For production applications (where writes are frequent), this is unlikely to be noticeable, but if youā€™re testing it out, you may need to manually make a write to see changes!

ā€‹
Limitations

ā€‹
Snapshot age

LiteFS Cloud stores ā€œsnapshotsā€ as a series of LTX files. For cost and performance reasons, LiteFS Cloud compacts these files into files containing transactions from five minute intervals. This means that when you select a timestamp for restoring, LiteFS Cloud takes the most recent file available from before that timestamp, which means that you may miss some transactions that happened within the five minute period before the timestamp you selected.

ā€‹
Snapshot restore requires database write

When you configure LiteFS Cloud, your LiteFS database considers LiteFS Cloud to be the authoritative source for the expected state of the database. When you restore a snapshot, LiteFS Cloud downloads the appropriate files, compacts them, and sends them to LiteFS the next time it asks for the state. However, currently LiteFS only asks LiteFS Cloud for the expected state on write, which means that you need to make a write before LiteFS updates the database state when you restore a snapshot.