Skip to content
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

Schema changes for application-level aggregation metrics. #3315

Merged
merged 2 commits into from
Jul 22, 2024
Merged

Schema changes for application-level aggregation metrics. #3315

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a84ed29
Schema changes for application-level aggregation metrics.
branlwyd Jul 22, 2024
855f260
Review.
branlwyd Jul 22, 2024
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
2 changes: 1 addition & 1 deletion aggregator_core/src/datastore.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,7 +102,7 @@ macro_rules! supported_schema_versions {
// version is seen, [`Datastore::new`] fails.
//
// Note that the latest supported version must be first in the list.
supported_schema_versions!(6, 5);
supported_schema_versions!(7, 6, 5);
branlwyd marked this conversation as resolved.
Show resolved Hide resolved

/// Datastore represents a datastore for Janus, with support for transactional reads and writes.
/// In practice, Datastore instances are currently backed by a PostgreSQL database.
Expand Down
1 change: 1 addition & 0 deletions db/00000000000007_task_aggregation_counters.down.sql
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
DROP TABLE task_aggregation_counters;
14 changes: 14 additions & 0 deletions db/00000000000007_task_aggregation_counters.up.sql
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,14 @@
-- Per-task report aggregation counters, used for metrics.
--
-- Fillfactor is lowered to improve the likelihood of heap-only tuple optimizations. See the
-- discussion around this setting for the task_upload_counters table.
CREATE TABLE task_aggregation_counters(
branlwyd marked this conversation as resolved.
Show resolved Hide resolved
id BIGINT GENERATED ALWAYS AS IDENTITY PRIMARY KEY, -- artificial ID, internal only
task_id BIGINT NOT NULL, -- task ID the counter is associated with
ord BIGINT NOT NULL, -- the ordinal index of the task aggregation counter

success BIGINT NOT NULL DEFAULT 0, -- reports successfully aggregated
Copy link
Contributor

Choose a reason for hiding this comment

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

Do we not want to track preparation failures by reason the way we do upload failures? If nothing else, keeping track of either total failures or total prep attempts would let us compute success rate.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

See response to inahga's comment:

I'm implementing the least amount of functionality that gives us what we need; I'm OK leaning on operational metrics (e.g. prometheus/grafana) & logging for other issues, at least until we have a justification for splitting them out. (And yeah, I think it would be easy to add more to this later.)

Success rate, specifically, can be tracked via the operational metrics.

Copy link
Contributor

Choose a reason for hiding this comment

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

I see we're not going to track the other failures that could come up, like task upload counters do. Rationale? I'm nominally OK with this, since if we want this functionality it should be fairly trivial to bolt it on later.

Copy link
Contributor Author

@branlwyd branlwyd Jul 22, 2024 •
edited
Loading

Choose a reason for hiding this comment

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

I'm implementing the least amount of functionality that gives us what we need; I'm OK leaning on operational metrics (e.g. prometheus/grafana) & logging for other issues, at least until we have a justification for splitting them out. (And yeah, I think it would be easy to add more to this later.)


CONSTRAINT task_aggregation_counters_unique_id UNIQUE(task_id, ord),
CONSTRAINT fk_task_id FOREIGN KEY(task_id) REFERENCES tasks(id) ON DELETE CASCADE
) WITH (fillfactor = 50);
Copy link
Contributor

Choose a reason for hiding this comment

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

Given that uploading these counter tables can be performance sensitive, I'm surprised we don't create an index on task ID. Maybe it's irrelevant because this table will be smallish anyway?

Copy link
Contributor

Choose a reason for hiding this comment

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

CONSTRAINT task_aggregation_counters_unique_id UNIQUE(task_id, ord) implicitly creates an index on (task_id, ord). We only ever INSERT on (task_id, ord) or SELECT task_id and aggregate over ord.

I have some query plans from task upload counters #2508 (comment).

Copy link
Contributor Author

Choose a reason for hiding this comment

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

UNIQUE constraints generate a (unique) index on the relevant fields, so we have an index on (task_id, ord).

See https://www.postgresql.org/docs/current/ddl-constraints.html#DDL-CONSTRAINTS-UNIQUE-CONSTRAINTS:

Adding a unique constraint will automatically create a unique B-tree index on the column or group of columns listed in the constraint.

Loading