feat(bigtable): route read_row/mutate_row through the accelerator with native fallback by mutianf · Pull Request #2 · mutianf/google-cloud-python · GitHub

Stack PR 2/7 — base: accel-01-daemon-subprocess-wrapper

Routes read_row and mutate_row through the accelerator daemon when eligible, via the method-name gate in _accelerator/_routing.py and per-method dispatch branches.

Accelerator stack

daemon subprocess wrapper
route read_row and mutate_row through daemon
forward auth/identity config to daemon
verify daemon identity before routing
authenticate RPCs with stdin handshake secret
pass Python client version for User-Agent
package prebuilt daemon into wheel

I'm just starting to get my head around this, but let some comments and questions from my first pass

This would work for grpc.RpcError exceptions. But there's also api_core.exceptions.GoogleAPICallError errors, which store this as grpc_status_code.

I don't know if that'll be relevant here

(We have similar code here. Maybe we should make this a shared helper?)

This could also live in google.cloud.bigtable.data.exceptions, depending on the usage

Do we want to prevent this fall-back if explicit is set?

Do we still need to collect the metrics in Python? Or does the accelerator handle that too?

I might need to double-check that read_rows still works as expected without the tracked_retry wrapper, unless you already verified that

Good point, I think this line should be removed.

Is read_rows_stream out of scope for now? That's the main part that I'd be worried about, since we have to be careful about how generators are wrapped/errors are propagated

Yeah, it's out of scope in this iteration.

I wonder if there's a solution here using decorators?

I like the idea of being able to mark each method with @accelerator_route('mutate_rows_via_accelerator'), and then have generic wrapper logic that handles the routing/fallback, with the native python implementation left mostly unchanged. That feels like it could give us some nice abstractions

But maybe that would introduce too many complications

This feels like a bigger change, maybe i'll do it in a future refactor.

Could it make sense to move this logic into accelerator_client.read_rows, to keep the accelerator implementation in one place?

accelerator implements bigtable.proto which doesn't have a read_row rpc. This already routes the request to accelerator_client.read_rows?

…h native fallback Change-Id: I6fc1d2563ca3185b12efb87b96eff2b7be438405

…back comments Change-Id: I191c52745e0cd10e5ab99e5e06135cd2e518bb9f

Change-Id: I72d5e456cdf722498ace021cc1f976cc1469a4dc

…NIMPLEMENTED Change-Id: Ifad0d9eb9d7b4d03fff07de2549a6d3294cf2f16

…lback Change-Id: I1b7382f6bfd8e70e7c6035e4b8ebb93ee3e282f3

mutianf commented Aug 10, 2026

View reviewed changes

Comment thread packages/google-cloud-bigtable/google/cloud/bigtable/data/_async/client.py Outdated

Comment thread packages/google-cloud-bigtable/google/cloud/bigtable/data/_async/client.py Outdated

mutianf force-pushed the accel-02-route-read-mutate branch from 934f2cd to b86802c Compare August 10, 2026 16:38

mutianf commented Aug 10, 2026

View reviewed changes

Comment thread packages/google-cloud-bigtable/google/cloud/bigtable/data/_async/client.py Outdated

mutianf force-pushed the accel-02-route-read-mutate branch 2 times, most recently from 8543853 to 86edefc Compare August 11, 2026 14:07

mutianf changed the title feat(bigtable): route read_row and mutate_row through accelerator daemon feat(bigtable): route read_row/mutate_row through the accelerator with native fallback Aug 11, 2026

mutianf commented Aug 11, 2026

View reviewed changes

mutianf force-pushed the accel-02-route-read-mutate branch 2 times, most recently from 1fc9db4 to 6d4ace5 Compare August 11, 2026 21:47

daniel-sanche reviewed Aug 21, 2026

View reviewed changes

mutianf added 5 commits August 24, 2026 14:57

feat(bigtable): route read_row/mutate_row through the accelerator wit…

328b432

…h native fallback Change-Id: I6fc1d2563ca3185b12efb87b96eff2b7be438405

docs(bigtable): drop stale Go-client references from accelerator fall…

c2ecd7e

…back comments Change-Id: I191c52745e0cd10e5ab99e5e06135cd2e518bb9f

chore(bigtable): add TODO to emit a metric in accelerator error handling

8fe1a9f

Change-Id: I72d5e456cdf722498ace021cc1f976cc1469a4dc

fix(bigtable): fall back to native immediately on first accelerator U…

47a7d79

…NIMPLEMENTED Change-Id: Ifad0d9eb9d7b4d03fff07de2549a6d3294cf2f16

fix(bigtable): address review feedback on accelerator routing and fal…

4419ba3

…lback Change-Id: I1b7382f6bfd8e70e7c6035e4b8ebb93ee3e282f3

mutianf force-pushed the accel-02-route-read-mutate branch from 1b2b842 to 4419ba3 Compare August 24, 2026 14:59



		def _grpc_code(exc: BaseException) -> StatusCode \| None:
		"""Best-effort extraction of a gRPC status code from an exception."""

+                  Never escapes the Table method that raises it: the method catches it and
+                  falls through to the native code path.
+                  """

+                              f"to the native client: {exc}",
+                              RuntimeWarning,
+                              stacklevel=2,
+                          )

+                          metric=ActiveOperationMetric(OperationType.READ_ROWS, is_streaming=False),
+                          retryable_exceptions=(),
+                      )
+                      row_merger._operation_metric.start_attempt()

+                              row_merger.request, timeout=operation_timeout
+                          )
+                          chunked_stream = row_merger.chunk_stream(stream)
+                          results = [a async for a in row_merger.merge_rows(chunked_stream)]

+                              )
+                          except _AcceleratorFallback:
+                              # Daemon can't serve this call; fall through to the native path.
+                              pass

                       return [row async for row in row_generator]
+                  @CrossSync.convert
+                  async def _read_row_via_accelerator(

feat(bigtable): route read_row/mutate_row through the accelerator with native fallback by mutianf · Pull Request #2 · mutianf/google-cloud-python · GitHub

feat(bigtable): route read_row/mutate_row through the accelerator with native fallback - #2

Conversation

mutianf commented Aug 7, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

daniel-sanche left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

daniel-sanche Aug 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

daniel-sanche Aug 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

daniel-sanche Aug 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

daniel-sanche left a comment •
edited
Loading

daniel-sanche Aug 21, 2026 •
edited
Loading

daniel-sanche Aug 21, 2026 •
edited
Loading

daniel-sanche Aug 21, 2026 •
edited
Loading