* Initial envoy cgroup resource monitor implementation
* Add cgroupv1 support; add metrics instrumentation
* Slight refactor for more efficient memory limit detection
Instead of reading memory.max/limit_in_bytes on every tick, we
read it once, then again only when it is modified.
To support this change, logic for computing the saturation was moved out
of the cgroup driver and into the resource monitor, and the driver
interface now has separate methods for reading memory usage and limit.
* Code cleanup/lint fixes
* Add platform build tags
* Add unit tests
* Fix lint issues
* Add runtime flag to allow disabling resource monitor
* Clamp saturation values to the range [0.0, 1.0]
* Switch to x/sys/unix; handle inotify IN_IGNORED events
If a request is made to the Pomerium metrics endpoint with an Accept
header requesting the Prometheus protobuf exposition format, some
metrics will be missing from the response.
These missing metrics are obtained by replaying the incoming request to
an OpenCensus metrics exporter. This exporter honors the request for the
protobuf format, however Pomerium expects this response to be in the
text format.
We can avoid this mismatch by explicitly requesting the text format from
the OpenCensus exporter, regardless of the incoming request's Accept
header.
(Note: the Pomerium metrics endpoint always responds with text format
metrics, even if the protobuf format is requested.)
* config: add enable_google_cloud_serverless_authentication to config protobuf
* use dependency injection for embedded envoy provider
* Revert "use dependency injection for embedded envoy provider"
This reverts commit 5c08990501.
* config: attach envoy version to Config to avoid metrics depending on envoy/files
* refactor backend, implement encrypted store
* refactor in-memory store
* wip
* wip
* wip
* add syncer test
* fix redis expiry
* fix linting issues
* fix test by skipping non-config records
* fix backoff import
* fix init issues
* fix query
* wait for initial sync before starting directory sync
* add type to SyncLatest
* add more log messages, fix deadlock in in-memory store, always return server version from SyncLatest
* update sync types and tests
* add redis tests
* skip macos in github actions
* add comments to proto
* split getBackend into separate methods
* handle errors in initVersion
* return different error for not found vs other errors in get
* use exponential backoff for redis transaction retry
* rename raw to result
* use context instead of close channel
* store type urls as constants in databroker
* use timestampb instead of ptypes
* fix group merging not waiting
* change locked names
* update GetAll to return latest record version
* add method to grpcutil to get the type url for a protobuf type
- gofumpt everything
- fix TLS MinVersion to be at least 1.2
- add octal syntax
- remove newlines
- fix potential decompression bomb in ecjson
- remove implicit memory aliasing in for loops.
Signed-off-by: Bobby DeSimone <bobbydesimone@gmail.com>
* metrics: support dynamic configuration settings
* add test
* trace: update configuration when settings change
* config: allow logging options to be configured when settings change
* envoy: allow changing log settings
* fix unexpected doc change
* fix tests
* pick a port at random
* update based on review
Document that metricRegistry is not safe for concurrently use. While at
it, remove t.Parallel() in tests which use metricRegistry, which causes
data race, caught by:
go test -race ./internal/telemetry/metrics
- Update envoy bootstrap config to protobufs
- Reorganize tracing config to avoid cyclic import
- Push down zipkin config to Envoy
- Update tracing options to provide sample rate
