Skip to content

[Prism] Add previousInput watermark and use it in bundleReady #36137

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

Merged
shunping merged 2 commits into from
Sep 13, 2025
Merged

[Prism] Add previousInput watermark and use it in bundleReady #36137

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
471708b
Add preRefreshedInput watermark and use it in bundleReady.
shunping Sep 12, 2025
17cad7d
Address the reviewer feedback.
shunping Sep 12, 2025
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
11 changes: 8 additions & 3 deletions sdks/go/pkg/beam/runners/prism/internal/engine/elementmanager.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1153,6 +1153,7 @@ type stageState struct {
input mtime.Time // input watermark for the parallel input.
output mtime.Time // Output watermark for the whole stage
estimatedOutput mtime.Time // Estimated watermark output from DoFns
previousInput mtime.Time // input watermark before the latest watermark refresh

pending elementHeap // pending input elements for this stage that are to be processesd
inprogress map[string]elements // inprogress elements by active bundles, keyed by bundle
Expand Down Expand Up @@ -2014,6 +2015,8 @@ func (ss *stageState) updateWatermarks(em *ElementManager) set[string] {
newIn = minPending
}

ss.previousInput = ss.input

// If bigger, advance the input watermark.
if newIn > ss.input {
ss.input = newIn
Expand Down Expand Up @@ -2171,11 +2174,13 @@ func (ss *stageState) bundleReady(em *ElementManager, emNow mtime.Time) (mtime.T
ptimeEventsReady := ss.processingTimeTimers.Peek() <= emNow || emNow == mtime.MaxTimestamp
injectedReady := len(ss.bundlesToInject) > 0

// If the upstream watermark and the input watermark are the same,
// then we can't yet process this stage.
// If the upstream watermark does not change, we can't yet process this stage.
// To check whether upstream water is unchanged, we evaluate if the input watermark, and
// the input watermark before the latest refresh are the same.
inputW := ss.input
_, upstreamW := ss.UpstreamWatermark()
if inputW == upstreamW {
previousInputW := ss.previousInput
if inputW == upstreamW && previousInputW == inputW {
slog.Debug("bundleReady: unchanged upstream watermark",
slog.String("stage", ss.ID),
slog.Group("watermark",
Expand Down
Loading