feat(storage): log tx opening location

[shekhirin]

To be able to catch the leaked transactions, we need to know their origin. Having a caller tracked up to the provider and the transaction ID should be enough.

2024-01-08T17:03:06.811863Z DEBUG storage::db::mdbx: Transaction opened caller=/Users/shekhirin/Projects/reth/crates/storage/provider/src/providers/database/mod.rs:250:14 id=305 mode=read-only
2024-01-08T17:03:06.817450Z  INFO reth::commands::node::events: Executing stage pipeline_stages=3/13 stage=Bodies checkpoint=222000 target=695895 stage_progress=31.90%
2024-01-08T17:03:06.817483Z DEBUG storage::db::mdbx: Transaction opened caller=/Users/shekhirin/Projects/reth/crates/stages/src/pipeline/mod.rs:351:53 id=306 mode=read-write
2024-01-08T17:03:07.006589Z  INFO reth::commands::node::events: pipeline_stages=3/13 stage=Bodies checkpoint=223000 target=695895 stage_progress=32.05% Stage committed progress
2024-01-08T17:03:07.020313Z DEBUG storage::db::mdbx: Transaction opened caller=/Users/shekhirin/Projects/reth/crates/storage/provider/src/providers/database/mod.rs:432:14 id=306 mode=read-only

---

The runtime cost of #[track_caller] should be negligible. No stack walk happens during the caller retrieval, and instead the caller is passed as an additional argument to the function. See https://rustc-dev-guide.rust-lang.org/backend/implicit-caller-location.html#generating-code-for-track_caller-callees for more information.

[mattsse]

lgtm

[gakonst]

send

[shekhirin]

concerned about slowdown during Execution stage, I see a lot of logs about opening new read-only transactions. Investigating.

[shekhirin]

okay, false alarm, they originated in RPC requests coming from Teku