-
Notifications
You must be signed in to change notification settings - Fork 0
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.
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
Join vertex - temporary PR on my own fork #1
Conversation
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
…ishing to the same buffer Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelmani <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
@@ -261,7 +261,7 @@ func Test_batchSyncWithMaxBatchSize(t *testing.T) { | |||
assert.NoError(t, err) | |||
err = wal.Write(&message) | |||
assert.NoError(t, err) | |||
assert.Equal(t, int64(222), tempWAL.prevSyncedWOffset) | |||
assert.Equal(t, int64(252), tempWAL.prevSyncedWOffset) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
because the ID changed to include vertex name, partition, etc, it made the overall message longer.
@@ -137,7 +137,7 @@ func (sp *SourceProcessor) Start(ctx context.Context) error { | |||
shuffleFuncMap := make(map[string]*shuffle.Shuffle) | |||
for _, edge := range sp.VertexInstance.Vertex.Spec.ToEdges { | |||
if edge.ToVertexType == dfv1.VertexTypeReduceUDF && edge.GetToVertexPartitionCount() > 1 { | |||
s := shuffle.NewShuffle(sp.VertexInstance.Vertex.GetName(), edge.GetToVertexPartitionCount()) | |||
s := shuffle.NewShuffle(edge.To, edge.GetToVertexPartitionCount()) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All Vertices publishing to a given Reduce Vertex need to be using the same hashing seed for distributing messages - thus, use the receiving Vertex name for that
Signed-off-by: Julie Vogelman <[email protected]>
// we only consider idle watermark if it is smaller than or equal to min of all the last processed watermarks. | ||
if headWMB.Watermark > e.getMinFromLastProcessed(headWMB.Watermark) { | ||
return wmb.WMB{} | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this needed to be moved into EdgeFetcherSet.GetHeadWMB()
Signed-off-by: Julie Vogelman <[email protected]>
@@ -67,7 +67,7 @@ metadata: | |||
spec: | |||
redis: | |||
native: | |||
version: 6.2.6` | |||
version: 7.0.11` |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
maybe should move this into a separate PR - just noticed it was outdated
containerTemplate: | ||
env: | ||
- name: NUMAFLOW_DEBUG | ||
value: "true" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
maybe should take out debug level environment variable for the sake of parity with the other tests
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Signed-off-by: Julie Vogelman <[email protected]>
Fixes numaproj#719
Enables Join Vertex (multiple Vertices publish to a single Vertex). This includes both cases where the Join Vertex is a map type, sink type, or Reduce type.
The implementation of this is based on this design document. (Sorry it's an internal link for now.)
Changes
The Design document goes over the need for an EdgeFetcherSet, which is the set of
EdgeFetchers
that a Vertex consumes Watermarks from.EdgeFetcherSet
traverses theEdgeFetchers
for each incoming Edge.In addition to that, some other changes made include:
Cycles
Plan is to enable Cycles in the pipeline as well as long as there is no Reduce Vertex at the point of the cycle or to the right (since Reduce vertex relies on the Watermark, and Cycles create a scenario in which late data arrives after Close-of-book). As of yet, Cycles have not been tested, and also validation needs to be added to prevent the Cycles mentioned above.