lncfg+scripts: use configurable global lock for postgres backends

Replace hardcoded WithGlobalLock assignment with configurable
options wallet postgres backends. Also add the WithGlobalLock
option to the channeldb table for postgres backends.

Defaults:
- channeldb: false (allow concurrent access)
- wallet: true (maintain safe single-writer behavior)

Users can now override these defaults via:
- db.postgres.channeldb-with-global-lock
- db.postgres.walletdb-with-global-lock

This gives operators flexibility while maintaining safe defaults
until full native SQL migration is complete.

Moreover exclude db.postgres.walletdb-with-global-lock check
in the sample config file script. We cannot easily check the
correct default because we set it later in the LND startup
sequence so we exclude it.
This commit is contained in:
ziggie committed 2025-11-28 09:22:19 +01:00
1 parent 2e16efed30
commit 4f051d97ea
4 files changed
+42 -10

No files matched your search

+10
View File
@@ -42,6 +42,16 @@ db.postgres.timeout=0
Connection timeout is disabled, to account for situations where the database
might be slow for unexpected reasons.
Moreover for particular kv tables we also add the option to access the
tables via a global lock (single wirter). This is a temorpary measure until
these particular tables have a native sql schema. This helps to mitigate
resource exhaustion in case LND experiencing high concurrent load:
* `db.postgres.walletdb-with-global-lock=true` to run LND with a single writer
for the walletdb_kv table (default is true).
* `db.postgres.channeldb-with-global-lock=false` to run the channeldb_kv table
with a single writer (default is false).
## Important note about replication
In case a replication architecture is planned, streaming replication should be avoided, as the master does not verify the replica is indeed identical, but it will only forward the edits queue, and let the slave catch up autonomously; synchronous mode, albeit slower, is paramount for `lnd` data integrity across the copies, as it will finalize writes only after the slave confirmed successful replication.