jobs: Eliminate job scheduler table scan for stats #78447
Assignees
Labels
A-jobs
C-enhancement
Solution expected to add code/behavior + preserve backward-compat (pg compat issues are exception)
T-jobs
Comments
miretskiy
added
C-enhancement
miretskiy
pushed a commit
to miretskiy/cockroach
that referenced
this issue
Mar 25, 2022
Remove `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job scheduler. These metrics are very expensive to compute as they involve running wider table scans against both `system.jobs` and `system.scheduled_job`. In addition to being expensive to compute, these metrics are not needed since the query can be executed directly if needed, and, in addition these metrics aref confusing since these metrics are per node, while the number of running jobs/schedules is cluster wide. More importantly, they can lead to job scheduler query being more expensive since they increase the read set of the scheduler transaction, thus causing txn restarts to be more expensive. Fixes cockroachdb#78447 Release Notes (enterprise): Remove expensive, unnecessary, and never used `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job schedulers. Release Justification: Stability fix for scheduled job system.
miretskiy
pushed a commit
to miretskiy/cockroach
that referenced
this issue
Mar 25, 2022
Remove `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job scheduler. These metrics are very expensive to compute as they involve running wider table scans against both `system.jobs` and `system.scheduled_job`. In addition to being expensive to compute, these metrics are not needed since the query can be executed directly if needed, and, in addition these metrics are confusing since these metrics are per node, while the number of running jobs/schedules is cluster wide. More importantly, they can lead to job scheduler query being more expensive since they increase the read set of the scheduler transaction, thus causing txn restarts to be more expensive. Fixes cockroachdb#78447 Release Notes (enterprise): Remove expensive, unnecessary, and never used `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job schedulers. Release Justification: Stability fix for scheduled job system.
miretskiy
pushed a commit
to miretskiy/cockroach
that referenced
this issue
Mar 25, 2022
Remove `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job scheduler. These metrics are very expensive to compute as they involve running wider table scans against both `system.jobs` and `system.scheduled_job`. In addition to being expensive to compute, these metrics are not needed since the query can be executed directly if needed, and, in addition these metrics are confusing since these metrics are per node, while the number of running jobs/schedules is cluster wide. More importantly, they can lead to job scheduler query being more expensive since they increase the read set of the scheduler transaction, thus causing txn restarts to be more expensive. Fixes cockroachdb#78447 Release Notes (enterprise): Remove expensive, unnecessary, and never used `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job schedulers. Release Justification: Stability fix for scheduled job system.
miretskiy
pushed a commit
to miretskiy/cockroach
that referenced
this issue
Mar 25, 2022
Remove `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job scheduler. These metrics are very expensive to compute as they involve running wider table scans against both `system.jobs` and `system.scheduled_job`. In addition to being expensive to compute, these metrics are not needed since the query can be executed directly if needed, and, in addition these metrics are confusing since these metrics are per node, while the number of running jobs/schedules is cluster wide. More importantly, they can lead to job scheduler query being more expensive since they increase the read set of the scheduler transaction, thus causing txn restarts to be more expensive. Fixes cockroachdb#78447 Release Notes (enterprise): Remove expensive, unnecessary, and never used `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job schedulers. Release Justification: Stability fix for scheduled job system.
blathers-crl bot
pushed a commit
that referenced
this issue
Mar 27, 2022
Remove `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job scheduler. These metrics are very expensive to compute as they involve running wider table scans against both `system.jobs` and `system.scheduled_job`. In addition to being expensive to compute, these metrics are not needed since the query can be executed directly if needed, and, in addition these metrics are confusing since these metrics are per node, while the number of running jobs/schedules is cluster wide. More importantly, they can lead to job scheduler query being more expensive since they increase the read set of the scheduler transaction, thus causing txn restarts to be more expensive. Fixes #78447 Release Notes (enterprise): Remove expensive, unnecessary, and never used `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job schedulers. Release Justification: Stability fix for scheduled job system.
miretskiy
pushed a commit
to miretskiy/cockroach
that referenced
this issue
Mar 28, 2022
Remove `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job scheduler. These metrics are very expensive to compute as they involve running wider table scans against both `system.jobs` and `system.scheduled_job`. In addition to being expensive to compute, these metrics are not needed since the query can be executed directly if needed, and, in addition these metrics are confusing since these metrics are per node, while the number of running jobs/schedules is cluster wide. More importantly, they can lead to job scheduler query being more expensive since they increase the read set of the scheduler transaction, thus causing txn restarts to be more expensive. Fixes cockroachdb#78447 Release Notes (enterprise): Remove expensive, unnecessary, and never used `schedules.round.schedules-ready-to-run` and `schedules.round.num-jobs-running` metrics from job schedulers. Release Justification: Stability fix for scheduled job system.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Scheduler creates loop stats for each iteration of its scheduling loop:
https://github.com/cockroachdb/cockroach/blob/master/pkg/jobs/job_scheduler.go#L201
This is bad because we are combing 2 table scans (jobs and schedules) in a same txn that is then used
to update scheduled job.
Such access patterns are very hard to serialize and can cause excessive intent buildup.
Jira issue: CRDB-14135
The text was updated successfully, but these errors were encountered: