cloudfoundry-community / firehose-to-syslog

Send firehose events from Cloud Foundry to syslog.
MIT License
44 stars 58 forks source link

Unable to Revert back to V1 loggregator due to performance concerns / And Cloud Foundry announcement #222

Open Scoobed opened 4 years ago

Scoobed commented 4 years ago

We received the following announcement around the Loggregator V2 RLP performance Concerns.

Based on the cancellation of the depreciation, we would like to revert back to V1 to see if fixes our issues around dropping logs. As Configured today, the client forces use of V2. And ignores the doppler-endpoint env variable.

https://github.com/cloudfoundry-community/firehose-to-syslog/blob/0324cba8f46364119e36353997ec630e70bee008/cli.go#L173

Fwd: [cf-dev] Cancellation of deprecation of Loggregator V1 Firehose

---------- Forwarded message --------- From: Jesse Weaver wjesse@vmware.com Date: Tue, Aug 11, 2020 at 3:36 PM Subject: [cf-dev] Cancellation of deprecation of Loggregator V1 Firehose To: cf-dev@lists.cloudfoundry.org cf-dev@lists.cloudfoundry.org

We in the Logging and Metrics team have decided to cancel the deprecation of the Loggregator V1 Firehose.

We had previously planned to deprecate the V1 then the V2 Firehose in favor of more modern, scalable APIs with more sustainable support paths (namely, logs over syslog and metrics over Prometheus endpoints). Difficulties encountered by downstream integrators, and concerns that have come up about the performance of the version of the V2 Firehose API accessible to outside integrations (1), have postponed the deprecation timeline significantly. This, combined with a shift in priority towards work on cf-for-k8s, has led us to believe that deprecation of the V1 Firehose is no longer the best path for operators, integrations, or ourselves.

  1. https://github.com/cloudfoundry/loggregator-release/issues/406 .,.,_