Skip to content

fix: Improve docs for min and max table offsets #14890

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

Merged
merged 3 commits into from
Nov 13, 2024
Merged

fix: Improve docs for min and max table offsets #14890

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
5db6214
Improve docs for min and max table offsets
salvacorts Nov 13, 2024
dd17922
Update docs/sources/shared/configuration.md
salvacorts Nov 13, 2024
2217467
update docs
salvacorts Nov 13, 2024
File filter

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
Improve docs for min and max table offsets
  • Loading branch information
@salvacorts
salvacorts committed Nov 13, 2024
commit 5db6214ea5d8986ab05738d5eb7920dd77246a7a
21 changes: 11 additions & 10 deletions docs/sources/shared/configuration.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1209,19 +1209,20 @@ planner:
# CLI flag: -bloom-build.planner.interval
[planning_interval: <duration> | default = 8h]

# Newest day-table offset (from today, inclusive) to build blooms for.
# Increase to lower cost by not re-writing data to object storage too
# frequently since recent data changes more often at the cost of not having
# blooms available as quickly.
# Newest day-table offset (from today, inclusive) to build blooms for. 0 start
# building from today, 1 from yesterday and so on. Increase to lower cost by
# not re-writing data to object storage too frequently since recent data
# changes more often at the cost of not having blooms available as quickly.
# CLI flag: -bloom-build.planner.min-table-offset
[min_table_offset: <int> | default = 1]
[min_table_offset: <int> | default = 0]

# Oldest day-table offset (from today, inclusive) to compact. This can be used
# to lower cost by not trying to compact older data which doesn't change. This
# can be optimized by aligning it with the maximum
# `reject_old_samples_max_age` setting of any tenant.
# Oldest day-table offset (from today, inclusive) to compact. 1 till
# yesterday, 2 till day before yesterday and so on. This can be used to lower
# cost by not trying to compact older data which doesn't change. This can be
# optimized by aligning it with the maximum `reject_old_samples_max_age`
# setting of any tenant.
# CLI flag: -bloom-build.planner.max-table-offset
[max_table_offset: <int> | default = 2]
[max_table_offset: <int> | default = 1]
Copy link
Contributor

Choose a reason for hiding this comment

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

Should we increase the default to 7? I think this would make sense in the broader context
Copy link
Contributor Author

Choose a reason for hiding this comment

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

I'm hesitant to do that until we get more confident about how good the build process scales. If data doesn't change as often, building today + tomorrow should work more most of the cases.
Copy link
Contributor

Choose a reason for hiding this comment

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

Ok with that

retention:
# Enable bloom retention.
Expand Down
4 changes: 2 additions & 2 deletions pkg/bloombuild/planner/config.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,14 +21,14 @@ type Config struct {
// RegisterFlagsWithPrefix registers flags for the bloom-planner configuration.
func (cfg *Config) RegisterFlagsWithPrefix(prefix string, f *flag.FlagSet) {
f.DurationVar(&cfg.PlanningInterval, prefix+".interval", 8*time.Hour, "Interval at which to re-run the bloom creation planning.")
f.IntVar(&cfg.MinTableOffset, prefix+".min-table-offset", 1, "Newest day-table offset (from today, inclusive) to build blooms for. Increase to lower cost by not re-writing data to object storage too frequently since recent data changes more often at the cost of not having blooms available as quickly.")
f.IntVar(&cfg.MinTableOffset, prefix+".min-table-offset", 0, "Newest day-table offset (from today, inclusive) to build blooms for. 0 start building from today, 1 from yesterday and so on. Increase to lower cost by not re-writing data to object storage too frequently since recent data changes more often at the cost of not having blooms available as quickly.")
// TODO(owen-d): ideally we'd set this per tenant based on their `reject_old_samples_max_age` setting,
// but due to how we need to discover tenants, we can't do that yet. Tenant+Period discovery is done by
// iterating the table periods in object storage and looking for tenants within that period.
// In order to have this done dynamically, we'd need to account for tenant specific overrides, which are also
// dynamically reloaded.
// I'm doing it the simple way for now.
f.IntVar(&cfg.MaxTableOffset, prefix+".max-table-offset", 2, "Oldest day-table offset (from today, inclusive) to compact. This can be used to lower cost by not trying to compact older data which doesn't change. This can be optimized by aligning it with the maximum `reject_old_samples_max_age` setting of any tenant.")
f.IntVar(&cfg.MaxTableOffset, prefix+".max-table-offset", 1, "Oldest day-table offset (from today, inclusive) to compact. 1 till yesterday, 2 till day before yesterday and so on. This can be used to lower cost by not trying to compact older data which doesn't change. This can be optimized by aligning it with the maximum `reject_old_samples_max_age` setting of any tenant.")
cfg.RetentionConfig.RegisterFlagsWithPrefix(prefix+".retention", f)
cfg.Queue.RegisterFlagsWithPrefix(prefix+".queue", f)
}
Expand Down
31 changes: 31 additions & 0 deletions pkg/bloombuild/planner/planner_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@ import (
"github.com/grafana/loki/v3/pkg/bloombuild/planner/queue"
"github.com/grafana/loki/v3/pkg/bloombuild/planner/strategies"
"github.com/grafana/loki/v3/pkg/bloombuild/protos"
iter "github.com/grafana/loki/v3/pkg/iter/v2"
"github.com/grafana/loki/v3/pkg/storage"
v1 "github.com/grafana/loki/v3/pkg/storage/bloom/v1"
"github.com/grafana/loki/v3/pkg/storage/chunk/cache"
Expand Down Expand Up @@ -606,6 +607,36 @@ func Test_deleteOutdatedMetas(t *testing.T) {
}
}

func TestMinMaxTables(t *testing.T) {
logger := log.NewNopLogger()
//logger := log.NewLogfmtLogger(os.Stdout)

cfg := Config{
PlanningInterval: 1 * time.Hour,
Queue: queue.Config{
MaxQueuedTasksPerTenant: 10000,
},
// From today till day before tomorrow
MinTableOffset: 0,
MaxTableOffset: 2,
}
planner := createPlanner(t, cfg, &fakeLimits{}, logger)

tables := planner.tables(time.Now())
require.Equal(t, 3, tables.TotalDays())

dayTables, err := iter.Collect(tables)
require.NoError(t, err)

todayTable := config.NewDayTable(config.NewDayTime(model.Now()), "index_")
yesterdayTable := config.NewDayTable(config.NewDayTime(model.Now().Add(-24*time.Hour)), "index_")
dayBeforeYesterdayTable := config.NewDayTable(config.NewDayTime(model.Now().Add(-48*time.Hour)), "index_")

require.Equal(t, dayBeforeYesterdayTable.Addr(), dayTables[0].Addr())
require.Equal(t, yesterdayTable.Addr(), dayTables[1].Addr())
require.Equal(t, todayTable.Addr(), dayTables[2].Addr())
}

type fakeBuilder struct {
mx sync.Mutex // Protects tasks and currTaskIdx.
id string
Expand Down
Loading