Release 0.13.0

Use jaeger storage integration helpers from upstream (#119)

* Use jaeger storage integration helpers from upstream

This commit deletes local copy of test helpers and fixtures and imports related
package from upstream jaeger as a golang pkg.

Signed-off-by: Arunprasad Rajkumar <[email protected]>

* Skip false positive golangci-linter error

Signed-off-by: Arunprasad Rajkumar <[email protected]>

Signed-off-by: Arunprasad Rajkumar <[email protected]>

Release 0.12.0

This release has several updates over the last few months:

• Adds support for a more efficient streaming grpc call for accepting spans from Jaeger proper
• The underlying Clickhouse library has been updated to 2.x
• Automated builds are using go 1.19 and now include arm64

Release 0.11.0

This release cleans up/deduplicates the internal SQL scripts used for creating tables, while adding two new optional features that can be enabled via YAML configuration. Existing users should not see any changes to behavior or schemas.

The new config options are:

• init_tables: An explicit option to indicate whether whether the plugin should automatically attempt to create tables in ClickHouse. This simplifies performing table initialization via a separate external process, or adding custom sql scripts to run in addition to the built-in ones. By default, this option replicates existing behavior, being enabled if init_sql_scripts_dir is empty, or disabled if init_sql_scripts_dir is provided.
• tenant: This allows running multiple separate Jaeger instances against a shared DB/tables. Existing users that don't specify a tenant should see no change to behavior.
  Specifying this option adds a tenant column to each of the tables, populating it with the provided value for all reads/writes. By default the partitioning is configured to be per-tenant.
  The existing per-DB multitenant option continues to be available as well, so you can mix-and-match both methods in the same ClickHouse instance as needed. However you cannot mix tenant and non-tenant instances in the same DB since they use slightly different schemas.
  For more information and config examples, see the updated multitenancy guide.

Release 0.10.0

When using replication, the operations materialized view was previously defaulting to a ReplicatedMergeTree. This release switches it to a SummingReplicatedMergeTree. Per the docs:

this significantly reduces storage volume and speeds up data selection

Existing replicated installations will remain on a ReplicatedMergeTree and would need to recreate or migrate the jaeger_operations table in order to pick up the change.

Non-replicated installations will not see any change as they are already using SummingMergeTree.

Release 0.9.0

• Add new configurable max_num_spans limit for read path: #101
• Fix storage of duplicate tags: #103
• Fixes to max_span_count handling: #104
  • Fix panic if limit is reached
  • Drop excess records (and complain via logs/metrics) if the limit is exceeded, rather than waiting: avoid stalling when limit is reached

Release 0.8.0

• Integrations tests from jaeger are added to package and to build workflow;
• Now operations are retrieved with SpanKind field. Also they are sorted by name;
• Span recording is now repeatable and asyncronous;
• Target Go version changed to 1.17.

Signed-off-by: Yury Frolov [email protected]

Release 0.7.0

Reflect package change

Signed-off-by: Pavol Loffay <[email protected]>

Release 0.6.0

Document that replication does not work on default databse (#66)

* Document that replication does not work on default databse

Signed-off-by: Pavol Loffay <[email protected]>

* Remove

Signed-off-by: Pavol Loffay <[email protected]>

* Remove

Signed-off-by: Pavol Loffay <[email protected]>

Release 0.5.1

Fix table name

Signed-off-by: Pavol Loffay <[email protected]>

Release 0.5.0

Make metrics private

Signed-off-by: Pavol Loffay <[email protected]>