Skip to content

Commit

Permalink
doc/BACKUP.md: Warn about version if using Litestream, sqlite3 `.du…
Browse files Browse the repository at this point in the history
…mp` is safe(r) now.
  • Loading branch information
ZmnSCPxj committed Oct 16, 2021
1 parent 697a8ca commit 2434e4d
Showing 1 changed file with 83 additions and 31 deletions.
114 changes: 83 additions & 31 deletions doc/BACKUP.md
Original file line number Diff line number Diff line change
Expand Up @@ -242,31 +242,6 @@ three or four storage devices.
BTRFS would probably work better if you were purchasing an entire set
of new storage devices to set up a new node.

## SQLite Litestream Replication
`/!\` WHO SHOULD DO THIS: Casual users

One of the simpler things on any system is to use Litestream to replicate the SQLite database.
It continuously streams SQLite changes to file or external storage - the cloud storage option
should not be used.
Backups/replication should not be on the same disk as the original SQLite DB.

/etc/litestream.yml :

dbs:
- path: /home/bitcoin/.lightning/bitcoin/lightningd.sqlite3
replicas:
- path: /media/storage/lightning_backup
and start the service using systemctl:

$ sudo systemctl start litestream

Restore:

$ litestream restore -o /media/storage/lightning_backup /home/bitcoin/restore_lightningd.sqlite3



## PostgreSQL Cluster

`/!\` WHO SHOULD DO THIS: Enterprise users, whales.
Expand Down Expand Up @@ -353,6 +328,59 @@ This can be difficult to create remote replicas due to the latency.

[pqsyncreplication]: https://www.postgresql.org/docs/13/warm-standby.html#SYNCHRONOUS-REPLICATION

## SQLite Litestream Replication
`/!\` WHO SHOULD DO THIS: Casual users

`/!\` **CAUTION** `/!\` This technique will only be safe on 0.10.2
or later.
Earlier versions will crash regularly with "database is locked" errors,
as Litestream puts a read-shared lock on the database.
0.10.2 adds a 60-second timeout for locking.

One of the simpler things on any system is to use Litestream to replicate the SQLite database.
It continuously streams SQLite changes to file or external storage - the cloud storage option
should not be used.
Backups/replication should not be on the same disk as the original SQLite DB.

You need to enable WAL mode on your database.
To do so, first stop `lightningd`, then:

$ sqlite3 lightningd.sqlite3
sqlite3> PRAGMA journal_mode = WAL;
sqlite3> .quit

Then just restart `lightningd`.

/etc/litestream.yml :

dbs:
- path: /home/bitcoin/.lightning/bitcoin/lightningd.sqlite3
replicas:
- path: /media/storage/lightning_backup

and start the service using systemctl:

$ sudo systemctl start litestream

Restore:

$ litestream restore -o /media/storage/lightning_backup /home/bitcoin/restore_lightningd.sqlite3

Because Litestream only copies small changes and not the entire
database (holding a read lock on the file while doing so), the
60-second timeout on locking should not be reached unless
something has made your backup medium very very slow.

Litestream has its own timer, so there is a tiny (but
non-negligible) probability that `lightningd` updates the
database, then irrevocably commits to the update by sending
revocation keys to the counterparty, and *then* your main
storage media crashes before Litestream can replicate the
update.
Treat this as a superior version of "Database File Backups"
section below and prefer recovering via other backup methods
first.

## Database File Backups

`/!\` WHO SHOULD DO THIS: Those who already have at least one of the
Expand Down Expand Up @@ -458,9 +486,33 @@ Even if the backup is not corrupted, take note that this backup
strategy should still be a last resort; recovery of all funds is
still not assured with this backup strategy.

You might be tempted to use `sqlite3` `.dump` or `VACUUM INTO`.
Unfortunately, these commands exclusive-lock the database.
A race condition between your `.dump` or `VACUUM INTO` and
`lightningd` accessing the database can cause `lightningd` to
crash, so you might as well just cleanly shut down `lightningd`
and copy the file at rest.
### `sqlite3` `.dump` or `VACUUM INTO` Commands

`/!\` **CAUTION** `/!\` This technique will only be safe on 0.10.2
or later.
Earlier versions will crash regularly with "database is locked"
errors, as `.dump` and `VACUUM INTO` put a read-shared lock on the
database.
0.10.2 adds a 60-second timeout for locking.

Use the `sqlite3` command on the `lightningd.sqlite3` file, and
feed it with `.dump "/path/to/backup.sqlite3"` or `VACUUM INTO
"/path/to/backup.sqlite3";`.

These create a snapshot copy that, unlike the previous technique,
is assuredly uncorrupted (barring any corruption caused by your
backup media).

However, if the copying process takes a long time (approaching the
timeout of 60 seconds) then you run the risk of `lightningd`
attempting to grab a write lock, waiting up to 60 seconds, and
then failing with a "database is locked" error.
Your backup system could `.dump` to a fast `tmpfs` RAMDISK or
local media, and *then* copy to the final backup media on a remote
system accessed via slow network, for example, to reduce this
risk.

It is recommended that you use `.dump` instead of `VACUUM INTO`,
as that is assuredly faster; you can just open the backup copy
in a new `sqlite3` session and `VACUUM;` to reduce the size
of the backup.

0 comments on commit 2434e4d

Please sign in to comment.