This is documentation for the next version of Tempo. For the latest stable release, go to the latest version.
Version 1.2 release notes
The Tempo team is excited to announce the release of Tempo 1.2. This release introduces Tempo search.
Features and enhancements
- Recent traces search The headline feature of Tempo 1.2 is the ability to search for data that is still in the ingester component. This experimental feature must be enabled using feature flags in both Grafana and Tempo. Refer to Tempo search for details.
- Scalable monolithic deployment mode (formerly scalable single binary deployment mode) The new scalable monolithic deployment mode is operationally simpler, although less robust than a fully distributed deployment. We consider it a balanced operational mode that some may find attractive.
- Massive performance improvements were achieved by more efficiently batching trace data being ingested. In terms of CPU needed, Tempo v1.2 is twice as efficient as Tempo v1.1. Refer to PR 1075 for details.
- API improvements Some informational endpoints were consolidated within the
status
endpoint. New endpoints were also added.
Upgrade considerations
When upgrading to Tempo v1.2, be aware of these changes:
- Support for v0 and v1 blocks were dropped as announced with Tempo v1.1. Refer to the Tempo v1.1 changelog for details.
- PR 1007 altered the Querier API. When upgrading components within a deployment, there will be a read outage until all queriers and query frontends have rolled to the upgraded version.
- API improvements consolidated informational endpoints within the
status
endpoint. - Metric
ingester_bytes_metric_total
is renamedingester_bytes_received_total
by PR 979. - Metric
cortex_runtime_config_last_reload_successful
is renamedtempo_runtime_config_last_reload_successful
by PR 945. - The
tempo-cli
flag--storage.trace.maintenance-cycle
is renamed-storage.trace.blocklist_poll
by PR 897.
Bug fixes
1.2.1 bug fixes
- Honor the default values of configuration parameters
max_bytes_per_trace
andmax_search_bytes_per_trace
. PR #1109 (@BitProcessor) - Reclassify the writing of empty traces to be a warning instead of an error. PR #1113 (@mdisibio)
1.2.0 bug fixes
- Fixed errors with magic numbers and other block mishandling when an ingester forcefully shuts down. Issue #937 (@mdisibio)
- Fixed a memory leak in the compactor component. PR #806 (@mdisibio)
- Set a span’s tag
span.kind
toclient
in query-frontend to allow query frontend spans to be paired with querier server spans. PR #975 (@mapno) - Metric
tempodb_backend_hedged_roundtrips_total
now correctly counts hedged roundtrips. PR #1079 (@joe-elliott) - Updated the
go-kit
logger package to remove spurious debug logs. PR #1094 (@bboreham)