diff --git a/docs/guides/administration/configuring-database.md b/docs/guides/administration/configuring-database.md index f1be0be..ffa2dff 100644 --- a/docs/guides/administration/configuring-database.md +++ b/docs/guides/administration/configuring-database.md @@ -157,6 +157,22 @@ This is particularly true when using HDDs instead of SSDs or NVME storage. ALTER SYSTEM SET (WAL_COMPRESSION = 'zstd'); ``` +### `jit` + +PostgreSQL's just-in-time (JIT) compilation can accelerate long-running, CPU-bound +analytical queries. It tends to hurt the short, latency-sensitive queries that +dominate Dependency-Track's workload. When the planner overestimates a query's cost, +it compiles machine code whose overhead can exceed the time it saves by orders of +magnitude. + +Dependency-Track already disables JIT for individual queries where it was found to be +actively harmful. Disabling it globally usually yields more consistent performance. +This optimization is common enough that PostgreSQL 19 changes the default to `off`. + +```sql +ALTER SYSTEM SET jit = off; +``` + ## Centralised connection pooling For large deployments (that is, upwards of 5 instances), it can become undesirable for diff --git a/docs/reference/configuration/database.md b/docs/reference/configuration/database.md index 50adff1..78c5363 100644 --- a/docs/reference/configuration/database.md +++ b/docs/reference/configuration/database.md @@ -107,6 +107,31 @@ deployments. For context on when and why to apply these, see the +### `jit` + +
| Default | +on |
+
|---|---|
| Recommendation | +off |
+
| References | ++ + | +