Skip to content

Conversation

@ibhatt-jumptrading
Copy link
Contributor

No description provided.

@github-actions
Copy link

Performance Measurements ⏳

Suite Baseline New Change
backtest mainnet-368528500-perf per slot 0.05087 s 0.050711 s -0.313%
backtest mainnet-368528500-perf snapshot load 1.644 s 1.618 s -1.582%
backtest mainnet-368528500-perf total elapsed 50.870417 s 50.711375 s -0.313%
firedancer mem usage with mainnet.toml 1005.23 GiB 1005.23 GiB 0.000%

@ibhatt-jumptrading ibhatt-jumptrading changed the title replay: adding better metrics [wip] replay: adding better metrics Dec 16, 2025
yufeng-jump
yufeng-jump previously approved these changes Dec 16, 2025

<counter name="SchedFull" summary="Times where sched is full and a FEC set can't be processed" />
<counter name="ReasmEmpty" summary="Times where reasm is empty and a FEC set can't be processed" />
<counter name="LeaderBidWait" summary="Times where replay is blocked by the the PoH tile not sending an end of leader message" />
Copy link
Contributor

Choose a reason for hiding this comment

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

the the

if( FD_UNLIKELY( (fec = fd_reasm_peek( ctx->reasm ))==NULL ) ) {
ctx->metrics.reasm_empty++;
ctx->metrics.reasm_latest_slot = ULONG_MAX;
ctx->metrics.reasm_latest_fec_idx = ULONG_MAX;
Copy link
Contributor

Choose a reason for hiding this comment

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

We might want these to monotonically increase, rather than resetting to ULONG_MAX. The way metrics make their way into the database, we might miss some high watermark values if we reset? I don't think we lose any information if we keep the reasm_latest_* values untouched in this branch, since reasm_empty will go up and that'll indicate an empty reasm.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

yeah agreed

@github-actions
Copy link

Performance Measurements ⏳

Suite Baseline New Change
backtest mainnet-368528500-perf per slot 0.050601 s 0.050511 s -0.178%
backtest mainnet-368528500-perf snapshot load 1.617 s 1.618 s 0.062%
backtest mainnet-368528500-perf total elapsed 50.601499 s 50.510731 s -0.179%
firedancer mem usage with mainnet.toml 1005.23 GiB 1005.23 GiB 0.000%

@ibhatt-jumptrading ibhatt-jumptrading merged commit d5cf53d into main Dec 16, 2025
13 checks passed
@ibhatt-jumptrading ibhatt-jumptrading deleted the ibhatt/replay_metrics branch December 16, 2025 09:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants