Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

trace_api_plugin enhancements #109

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

trace_api_plugin enhancements #109

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 20 additions & 0 deletions trace-api-plugin/trace_api_plugin_enhancements
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,20 @@
# Trace API Plugin enhancements

## ABI Support

Currently `trace_api_plugin` only supports decoding `action` and `return_value` data via command line provided `trace-rpc-abi` files. This requires manually
providing new ABI file any time the ABI changes.
Add using on-chain ABI at time of action execution for decoding of the `action` and `return_value` data.
Provide option to not track ABI if user does not require data expanded via ABI. ABI decoding should be the default.

## Index transactions by transaction id for quicker retrieval.

Currently `trace_api_plugin` does a sequential scan for `/v1/trace_api/get_transaction_trace`. Add an index for faster lookup.

## Investigate removing `fsync`

We do not use `fsync` for other file writing. Determine if `fflush` is sufficient.

## Use SQLite for storage

Usage would simplify the index on transaction issue and provide easier storage/lookup of ABI. Also solves the `fsync` question.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does it solve it or just move it? I thought (by default) a transaction in sqlite implies an fsync. So if, for example, we do an sqlite transaction for each block, it'd be no different in fsync overhead as before.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

https://www.sqlite.org/pragma.html#pragma_synchronous
By default fsync is still used. So move is a better description. It does look like it can be smarter about the fsync than our current implementation. Although, our current implementation likely doesn't even need it.

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If we use "journal_mode=WAL" and "synchronous=normal" this should reduce fsync to only when the WAL is checkpointing.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

From our 11/8/2022 conversation on this topic:

  • need to understand potential impacts on transaction processing time
  • might explore alternative proposal adding filtering capabilities to SHiP.