Queue-depth ODS counters (#5913)#5913
Closed
adityas-meta wants to merge 1 commit into
Closed
Conversation
adityas-meta
force-pushed
the
export-D108312495
branch
from
c491e9b to
e869df4
Compare
meta-codesync
Bot
changed the title
adityas-meta
added a commit
to adityas-meta/FBGEMM-1
that referenced
this pull request
Jun 17, 2026
Summary:
Add observability for the two queue depths in the RES streaming pipeline. (1) `weights_to_stream_queue_` (MPSC stream queue in fbgemm) emits `res.queue.stream_mpsc_depth` after every dequeue. (2) `tensor_to_dedup_queues_[i]` (per-shard dedup queues in training_ps) emit `res.queue.dedup_depth.shard_{i}` per shard plus a `res.queue.dedup_depth.total` rollup, bumped at the top of `co_stream_tensors`. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance.
Both surfaces use OBC via `TrainingPsOdsLogger::bumpKeyLatency` (emits AVG+P50+P99 — the right stats for depth percentiles), on the `raw_embedding_streaming` category. The handler already owned `ods_logger_`; the fbgemm-side `RawEmbeddingStreamer` reuses the `TrainingPsOdsLogger` member added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303 `ResGauges.h` helper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.)
Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on `RawEmbeddingStreamer`. Both feed the master observability initiative T269497764.
Design note: per-shard cardinality is N OBC keys x 3 stats; `num_dedup_threads_` is configurable but typically small (4-8); `.total` is for at-a-glance, per-shard for hot-spot debug.
Differential Revision: D108312495
adityas-meta
added a commit
to adityas-meta/FBGEMM-1
that referenced
this pull request
Jun 17, 2026
Summary: X-link: facebookresearch/FBGEMM#2832 Add observability for the two queue depths in the RES streaming pipeline. (1) `weights_to_stream_queue_` (MPSC stream queue in fbgemm) emits `res.queue.stream_mpsc_depth` after every dequeue. (2) `tensor_to_dedup_queues_[i]` (per-shard dedup queues in training_ps) emit `res.queue.dedup_depth.shard_{i}` per shard plus a `res.queue.dedup_depth.total` rollup, bumped at the top of `co_stream_tensors`. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance. Both surfaces use OBC via `TrainingPsOdsLogger::bumpKeyLatency` (emits AVG+P50+P99 — the right stats for depth percentiles), on the `raw_embedding_streaming` category. The handler already owned `ods_logger_`; the fbgemm-side `RawEmbeddingStreamer` reuses the `TrainingPsOdsLogger` member added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303 `ResGauges.h` helper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.) Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on `RawEmbeddingStreamer`. Both feed the master observability initiative T269497764. Design note: per-shard cardinality is N OBC keys x 3 stats; `num_dedup_threads_` is configurable but typically small (4-8); `.total` is for at-a-glance, per-shard for hot-spot debug. Differential Revision: D108312495
adityas-meta
force-pushed
the
export-D108312495
branch
from
e869df4 to
5c5978c
Compare
adityas-meta
added a commit
to adityas-meta/FBGEMM-1
that referenced
this pull request
Jun 17, 2026
Summary: X-link: facebookresearch/FBGEMM#2832 Add observability for the two queue depths in the RES streaming pipeline. (1) `weights_to_stream_queue_` (MPSC stream queue in fbgemm) emits `res.queue.stream_mpsc_depth` after every dequeue. (2) `tensor_to_dedup_queues_[i]` (per-shard dedup queues in training_ps) emit `res.queue.dedup_depth.shard_{i}` per shard plus a `res.queue.dedup_depth.total` rollup, bumped at the top of `co_stream_tensors`. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance. Both surfaces use OBC via `TrainingPsOdsLogger::bumpKeyLatency` (emits AVG+P50+P99 — the right stats for depth percentiles), on the `raw_embedding_streaming` category. The handler already owned `ods_logger_`; the fbgemm-side `RawEmbeddingStreamer` reuses the `TrainingPsOdsLogger` member added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303 `ResGauges.h` helper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.) Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on `RawEmbeddingStreamer`. Both feed the master observability initiative T269497764. Design note: per-shard cardinality is N OBC keys x 3 stats; `num_dedup_threads_` is configurable but typically small (4-8); `.total` is for at-a-glance, per-shard for hot-spot debug. Differential Revision: D108312495
adityas-meta
force-pushed
the
export-D108312495
branch
2 times, most recently
from
76587ba to
d28c754
Compare
adityas-meta
added a commit
to adityas-meta/FBGEMM-1
that referenced
this pull request
Jun 22, 2026
Summary: X-link: facebookresearch/FBGEMM#2832 Add observability for the two queue depths in the RES streaming pipeline. (1) `weights_to_stream_queue_` (MPSC stream queue in fbgemm) emits `stream_mpsc_depth` after every dequeue. (2) `tensor_to_dedup_queues_[i]` (per-shard dedup queues in training_ps) emit `dedup_depth.shard_{i}` per shard plus a `dedup_depth.total` rollup, bumped at the top of `co_stream_tensors`. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance. Both surfaces use OBC via `TrainingPsOdsLogger::bumpKeyGauge` (emits AVG+P50+P99 — the right stats for depth percentiles), on the `raw_embedding_streaming` category. The handler already owned `ods_logger_`; the fbgemm-side `RawEmbeddingStreamer` reuses the `TrainingPsOdsLogger` member added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303 `ResGauges.h` helper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.) Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on `RawEmbeddingStreamer`. Both feed the master observability initiative T269497764. Design note: per-shard cardinality is N OBC keys x 3 stats; `num_dedup_threads_` is configurable but typically small (4-8); `.total` is for at-a-glance, per-shard for hot-spot debug. Differential Revision: D108312495
adityas-meta
added a commit
to adityas-meta/FBGEMM-1
that referenced
this pull request
Jun 23, 2026
Summary: X-link: facebookresearch/FBGEMM#2832 Add observability for the two queue depths in the RES streaming pipeline. (1) `weights_to_stream_queue_` (MPSC stream queue in fbgemm) emits `stream_mpsc_depth` after every dequeue. (2) `tensor_to_dedup_queues_[i]` (per-shard dedup queues in training_ps) emit `dedup_depth.shard_{i}` per shard plus a `dedup_depth.total` rollup, bumped at the top of `co_stream_tensors`. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance. Both surfaces use OBC via `TrainingPsOdsLogger::bumpKeyGauge` (emits P50+P99 — the right stats for depth percentiles), on the `raw_embedding_streaming` category. The handler already owned `ods_logger_`; the fbgemm-side `RawEmbeddingStreamer` reuses the `TrainingPsOdsLogger` member added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303 `ResGauges.h` helper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.) Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on `RawEmbeddingStreamer`. Both feed the master observability initiative T269497764. Design note: per-shard cardinality is N OBC keys x 3 stats; `num_dedup_threads_` is configurable but typically small (4-8); `.total` is for at-a-glance, per-shard for hot-spot debug. Differential Revision: D108312495
adityas-meta
force-pushed
the
export-D108312495
branch
2 times, most recently
from
453b3b6 to
06e6872
Compare
adityas-meta
added a commit
to adityas-meta/FBGEMM-1
that referenced
this pull request
Jun 26, 2026
Summary: X-link: facebookresearch/FBGEMM#2832 Add observability for the two queue depths in the RES streaming pipeline. (1) `weights_to_stream_queue_` (MPSC stream queue in fbgemm) emits `stream_mpsc_depth` after every dequeue. (2) `tensor_to_dedup_queues_[i]` (per-shard dedup queues in training_ps) emit `dedup_depth.shard_{i}` per shard plus a `dedup_depth.total` rollup, bumped at the top of `co_stream_tensors`. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance. Both surfaces use OBC via `TrainingPsOdsLogger::bumpKeyGauge` (emits P50+P99 — the right stats for depth percentiles), on the `raw_embedding_streaming` category. The handler already owned `ods_logger_`; the fbgemm-side `RawEmbeddingStreamer` reuses the `TrainingPsOdsLogger` member added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303 `ResGauges.h` helper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.) Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on `RawEmbeddingStreamer`. Both feed the master observability initiative T269497764. Design note: per-shard cardinality is N OBC keys x 3 stats; `num_dedup_threads_` is configurable but typically small (4-8); `.total` is for at-a-glance, per-shard for hot-spot debug. Reviewed By: FriedCosey Differential Revision: D108312495
Summary: X-link: facebookresearch/FBGEMM#2832 Add observability for the two queue depths in the RES streaming pipeline. (1) `weights_to_stream_queue_` (MPSC stream queue in fbgemm) emits `stream_mpsc_depth` after every dequeue. (2) `tensor_to_dedup_queues_[i]` (per-shard dedup queues in training_ps) emit `dedup_depth.shard_{i}` per shard plus a `dedup_depth.total` rollup, bumped at the top of `co_stream_tensors`. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance. Both surfaces use OBC via `TrainingPsOdsLogger::bumpKeyGauge` (emits P50+P99 — the right stats for depth percentiles), on the `raw_embedding_streaming` category. The handler already owned `ods_logger_`; the fbgemm-side `RawEmbeddingStreamer` reuses the `TrainingPsOdsLogger` member added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303 `ResGauges.h` helper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.) Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on `RawEmbeddingStreamer`. Both feed the master observability initiative T269497764. Design note: per-shard cardinality is N OBC keys x 3 stats; `num_dedup_threads_` is configurable but typically small (4-8); `.total` is for at-a-glance, per-shard for hot-spot debug. Reviewed By: FriedCosey Differential Revision: D108312495
adityas-meta
force-pushed
the
export-D108312495
branch
from
06e6872 to
5a1ae0b
Compare
Contributor
|
This pull request has been merged in 969615c. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/2832
Add observability for the two queue depths in the RES streaming pipeline. (1)
weights_to_stream_queue_(MPSC stream queue in fbgemm) emitsstream_mpsc_depthafter every dequeue. (2)tensor_to_dedup_queues_[i](per-shard dedup queues in training_ps) emitdedup_depth.shard_{i}per shard plus adedup_depth.totalrollup, bumped at the top ofco_stream_tensors. Backpressure today is invisible until a trainer crashes; these signals let oncall see saturation trends and per-shard hot-spots in advance.Both surfaces use OBC via
TrainingPsOdsLogger::bumpKeyGauge(emits P50+P99 — the right stats for depth percentiles), on theraw_embedding_streamingcategory. The handler already ownedods_logger_; the fbgemm-sideRawEmbeddingStreamerreuses theTrainingPsOdsLoggermember added by the parent diff D107811590 (constructed only when streaming is enabled). No fb303: OBC reaches ODS via the host-level agent without per-process export config. (An earlier revision used an fb303ResGauges.hhelper for the fbgemm gauge; removed per review feedback on the sibling silent-failure diff.)Stacked on D107811590 (silent-failure ODS counters), which introduces the shared OBC logger member on
RawEmbeddingStreamer. Both feed the master observability initiative T269497764.Design note: per-shard cardinality is N OBC keys x 3 stats;
num_dedup_threads_is configurable but typically small (4-8);.totalis for at-a-glance, per-shard for hot-spot debug.Reviewed By: FriedCosey
Differential Revision: D108312495