| feat(all): Add release version to binaries (#6912) * set verions on Go services * fix tags * set tags on all components * fix kotlin version * remove auto-gen file * lint * copyright * Update operator/Makefile Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> * Update scheduler/data-flow/Makefile Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> * Update scheduler/Makefile Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> * log version with logger --------- Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> | 10 months ago |
| GitBook: No commit message | 1 year ago |
| feat(e2e): test for changing models and model/server auto-scaling (#7074) * test for changing models and model/server auto-scaling * server wait for status change * copyright | 9 months ago |
| fix: add PipelineFailedTerminating in PipelineStatus enum in scheduler protos (#7080) | 9 months ago |
| fix(controller): Server scaling spec (#6613) * fix Server spec validation forcing expected replicas to equal min replicas * use interface for scheduler client to enable de-coupling and enable writing of tests. Add tests cases to integration test suite * TODO * lint * show lint errors * fix lint * lint * lint * lint * run go generate on PR to ensure mocks are kept aligned with interfaces * fix tests * lint * go generate * fix race * return concrete type instead of interface * enforce copyright is up-to-date in PRs * naming | 1 year ago |
| Rename 'Server Runtime' to 'Seldon Runtime' (#7085) * Rename 'Server Runtime' to 'Seldon Runtime' * Fix formatting and punctuation in seldonruntime.md | 9 months ago |
| Update faqs.md (#6515) * Update faqs.md * Update faq - old docs | 1 year ago |
| Merge GitBook Docs with V2 master (#5943) **PR Summary: New Documentation Site for Seldon Core v2 (#5760)** This PR introduces a comprehensive update to the documentation site for Seldon Core v2, aligning the format with GitBook requirements. Key changes include: - **File and Directory Restructuring**: Moved and renamed multiple files and directories to match GitBook's expected format, including APIs, models, examples, CLI, Kubernetes, and performance tests sections. - **Content Updates**: - Completed the APIs section and Getting Started guide. - Enhanced and reorganized the models directory and various other docs. - Outliers, drift, servers, pipelines, inference, explainers, and experiments docs moved to the root directory with improvements. - New documentation for HPA-based autoscaling for scaling both Models and Servers, referencing Model RPS metrics from Prometheus. - Added per-component labels and annotations in documentation. - **Images**: Consolidated images into a dedicated directory and updated links throughout the docs. - **Format Fixes**: Resolved broken links, corrected typos, and fixed references in Kubernetes and other sections. - **GitBook-Specific Updates**: - Updated `SUMMARY.md` for GitBook compatibility. - Reformatted the development, metrics, and K8s directories for GitBook’s markdown flavor. - Added GitHub Gist references for certain files (replacing hard-coded references). Co-authored by multiple contributors from the Seldon Core team. @Rajakavitha1 and @lc525 | 2 years ago |
| fix(controller): Server scaling spec (#6613) * fix Server spec validation forcing expected replicas to equal min replicas * use interface for scheduler client to enable de-coupling and enable writing of tests. Add tests cases to integration test suite * TODO * lint * show lint errors * fix lint * lint * lint * lint * run go generate on PR to ensure mocks are kept aligned with interfaces * fix tests * lint * go generate * fix race * return concrete type instead of interface * enforce copyright is up-to-date in PRs * naming | 1 year ago |
| fix(mlserver): turn off logging request logs by default (#7042) * turn off logging request logs by default * remove accidental generated manifests * revert * copyright | 10 months ago |
| fix(ansible): Upgrade deps in ansible install (#6146) * update to mlserver 1.6.1 * upgrade strimzi operator * upgrade kafka to 3.8 * smoke test fixes * stress test fixes * remove kraft feature gate * add kafka strimiz (raft only) * upgrade prometheus * upgrade otel collector to 0.114.1 * upgrade jaeger to 1.62.0 * upgrade grafana to 11.3.1 * install prometheus adapter 0.12.0 * remove/add extra line * changes to kafka per review comment * remove seldon from prometheus adapter name * add hpa configmap | 1 year ago |
| Re-generate license info (#6575) Co-authored-by: lc525 <19005+lc525@users.noreply.github.com> | 1 year ago |
| fix: revert to previous directory where generated outputted CRDs are saved (#7060) | 10 months ago |
| Add files via upload (#6934) | 11 months ago |
| GitBook: No commit message | 1 year ago |
| feat(all): Add release version to binaries (#6912) * set verions on Go services * fix tags * set tags on all components * fix kotlin version * remove auto-gen file * lint * copyright * Update operator/Makefile Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> * Update scheduler/data-flow/Makefile Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> * Update scheduler/Makefile Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> * log version with logger --------- Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> | 10 months ago |
| Docs update (#150) * update docs * lint * Add API docs * further docs * add local examples * add k8s examples * rewrite about | 4 years ago |
| feat(e2e-tests): all suites run in one test (#7082) | 9 months ago |
| feat(tests): pipeline scalability tests (#6813) * wip: pipeline scalability tests * pipeline scalability tests wip * wip: load test 2 pipelines * skip setup * fix skip * 1 node pipeline test w/o scaled svcs to establish base line * make kind cluster name configurable * debug * ansible: deploy clickhouse and wireup OTEL to export data to it * fix naming * name too long * fix endpoint * add ability to set request ID to assist with tracing * trace pipeline req * fix ansible var * fix tracing for req id * use prefix for req id, as req id must be unique * add req id to grpc trace * make naming more unique * fix missing trace * fix ctx * revert - not working * record trace for how long it took a model-gw worker to pick up job from channel * scale model gw test * test model via agent * fix agent grpc, need correct model generation version * scale servers to 5 * increase retries for when increasing replica count by large amount * log result * fix payload * grpc keep-alives * fix panic in shutdown of model-gw * shutdown consumers concurrently to speed up shutdown phase and not hit the k8s deadline * wip: more pipeline tests * stop flooding of logs * nit * parallel pipeline test working * fix tests * scale services up/down during test * more scalbility tests * nit * scale servers to 5 * replica model to all servers * increase timeout * use multiple gRPC conns from model-gw * use gRPC * make configurable * more config options * sleep to fix race * missing import * comment out * wait for server to be ready * fix got string instead of int issue * make args consistent * typo * fix server blocked on draining * ability for tests to set memory limits on dataflow-engine * make model gw num of workers configurable and dataflow engine memory limits * clean up * nit * revert changes to use in different user stories * nit * fix ClickHouse installation * clean up tests * debug * fix merge rebase issues * wip * fix tests * docker job tests * nit * lint * PR comments | 1 year ago |
| feat(kafka): Add support for SASL OAUTHBEARER mechanism for Kafka (#5127) * Add OAuth option to supported SASL mechanisms * Use string templates instead of concatenation * Rename var for clarity * Refactor SASL handling & add structure for OAuth support This includes includes the values which need to be set, but without providing the client ID and secret, and so on. This has yet to be done. * Add SASL OAuth CLI arg * Add k8s secrets provider This will replace the existing one, splitting the responsibilities between retrieving secrets and interpreting their contents. This will enable the upcoming SASL OAuth secret handler to reuse the underlying k8s secret provider. This implementation is also simpler than before and provides logging. * Add interface for secrets provider * Refactor SASL password provider to use k8s secrets provider * Add default object for SASL password provider This provides flexibility around testing while maintaining the convenience of a readily-available object. * Update use of SASL password provider in Kafka config * Add SASL OAuth config provider * Move logger to companion object for SASL password provider * Refactor SASL OAuth provider for consistent levels of abstraction * Add default object for SASL OAuth provider This is for the same reason as for the SASL password provider. This is a lazy instantiation. * Change SASL password provider to use lazy instantiation * Use SASL OAuth config in Kafka properties creation * Rename var for clarity * Rename field in data class for disambiguation * Add OAuth secret field to SASL config * Use OAuth secret for SASL OAuth provider * Remove unused import * Only use SASL password from k8s if using password-based auth * Add Kafka SASL OAuth secret in Helm chart & plain k8s manifest * Add missing copyright notices * Decode OAuth config values from secret as base-64 byte arrays * Use URL-safe base64 decoder * Fix conversion of byte arrays to strings for OAuth config values * Use Kotlin-esque string builder construct * Wrap JAAS module option values in double quotes & terminate it with a semi-colon * Fix parsing of SASL/OAuth extensions from secret for robustness * Fix wrong key for SASL/OAuth extensions in secret * Formatting * Remove unused type extension for decoding Base64 byte arrays * Formatting for legibility * Fix wording in log lines referencing wrong type of data * Use secret name in logging for SASL password retrieval * Update param name & type for clarity on semantics/usage * Clarify wording in log messages * Use single SASL client secret instead of separate OAuth & password secrets in dataflow engine * Ensure newlines at the end of files * Remove unused data type * Use sealed classes to represent config for SASL mechanisms This ensures only the relevant config values can be used in any given logic. It also allows for validation in the future when creating configs, e.g. to reject missing parameters. * Simplify creation of JAAS config when using SASL password mechanisms * Remove OAuth secret config for k8s deployment mechanisms (Helm & YAML manifest) * Change default Helm value for Kafka SASL secret password path This new default is in line with the Kubernetes setup, as provided by Helm, which makes use of a secret rather than a directly-mounted file. * add helper target for go modules * initial OAUTH configuration code for Kafka * fix missing port setting; * fix prefix of some global variables * temporarily fix lint * rename tls.go to auth.go * move oauth package under kafka config * add test * lint * go mod tidy * update Kafka from 1.9.1 to 1.9.2 * add scope support for kafka oauth protocol * reuse secret variable set by security.kafka.sasl.client.secret * remove debug overwrite * add documentation * minor fixes * Update docs/source/contents/getting-started/kubernetes-installation/security/confluent-oauth.md * make scope generic here * leave scope as empty to be set only when required * missing yaml changes * add comment * docs fix * Update docs/source/contents/kubernetes/kafka/index.md Co-authored-by: Alex Rakowski <20504869+agrski@users.noreply.github.com> --------- Co-authored-by: Alex Rakowski <20504869+agrski@users.noreply.github.com> | 3 years ago |
| feat: Add kube api for k6 scaffolding (#5494) * make tfsimple_pipeline test create random models underneath * update alpine base image and install xk6 * make namespace configurable in shell script * add xk6 to .gitignore * add xk6 install as make target * adjust wrapper to use k6 with xk6 * add note in readme regarding xk6 * add k8s test script scenario * adjust namespace variable * ability to set namespace in kustomize * update model script * add newline in file * add NAMESPACE as envar * wire up namespace * parse pod namespace from envar * use NAMESPACE env var * add service account and remove unnecessary secret * have main scenario back as default * review comments | 2 years ago |
| enable Lychee for docs on v2 (#6786) * enable Lychee for Core 1 docs on v2 * enable Lychee for Core 1 docs on v2 * fix: correct glob pattern in lychee link checker workflow * fix: Optimize lychee link checker workflow * fix: Optimize lychee link checker workflow * fix: Update the read.me * fix: Update the gb-docs and read.me * fix:updated a file outside gb docs * fix:updated a file outside gb docs * fix:updated a file outside gb docs * chore: restore missing empty line in docs-gb README * fix: remove extra line from gb-docs workflow | 1 year ago |
| Add ReadTheDocs config file (#663) This is generally good for informing RTD on how to behave, but is also required ahead of the migration to the SCv1 repo. | 3 years ago |
| Update Changelog (#6605) Co-authored-by: lc525 <19005+lc525@users.noreply.github.com> | 1 year ago |
| Update CONTRIBUTING.md to add notice on licensing changes and links to previous contributions (#5555) | 2 years ago |
| feat: LPL release - refactor(kafka): Refactor Kafka SASL logic - fix(dataflow): Fix for batching does not support raw tensors - docs(metrics): [V2D-1220] Update documentation for podmonitor changes - refactor(kafka): [SCV2-32] Refactor Kafka SASL password logic for consistency - refactor(kafka): [SCV2-61] Restrict scope of locks in Kafka auth setup - fix(dataflow): [SCV2-18] Fix for batching does not handle mixed raw and normal tensors - fix(auth): [SCV2-60] Add checks for empty/whitespace namespaces to Kafka auth stores - build(operator): [SCV2-53] Fix operator CVEs due to stale Go packages - build(hodometer): [SCV2-53] Fix Hodometer CVEs due to stale Go packages - build(envoy): [SCV2-53] Update Envoy base image from UBI 9.1 to 9.3 - build(hodometer): [SCV2-53] Update Hodometer base images from UBI 9.2 to 9.3 - build(rclone): [SCV2-53] Update Rclone base image to UBI-9.3 - build(dataflow): [SCV2-53] Update dataflow engine base image from UBI9 OpenJDK 1.14 to 1.16 - build(operator): [SCV2-53] Update operator base image from UBI 9.2 to 9.3 - build(agent): [SCV2-53] Update agent base image from UBI 9.2 to 9.3 - build(scheduler): [SCV2-53] Update scheduler base image from UBI 9.2 to 9.3 - build(model-gateway): [SCV2-53] Update model gateway base image from UBI 9.2 to 9.3 - build(pipeline-gateway): [SCV2-53] Update pipeline gateway base image from UBI 9.2 to 9.3 - build(runtime-components): [SCV2-53] Update stale deps in scheduler module - fix: CVEs by upgrading to g1.20 - fix: Upgrade rclone and grafana - fix: LPL changes - fix(scheduler): fix scheduler segfault when agent 0 disconnects - fix(agent): Rclone config resync - fix(scheduler): non latest models consistency when server disconnects - feature: configuration of images using private registries via ansible - fix(dataflow): deadlock when Kafka stream never reaches 'running' state - fix: Update [CONTRIBUTING.md](http://contributing.md/) to remove reference to open source license --------- Co-authored-by: Alex Rakowski <20504869+agrski@users.noreply.github.com> Co-authored-by: Michael Cheah <20640150+michaelcheah@users.noreply.github.com> Co-authored-by: Lucian Carata <lc525@users.noreply.github.com> Co-authored-by: paulb-seldon <141156400+paulb-seldon@users.noreply.github.com> | 2 years ago |
| fix(controller): Server scaling spec (#6613) * fix Server spec validation forcing expected replicas to equal min replicas * use interface for scheduler client to enable de-coupling and enable writing of tests. Add tests cases to integration test suite * TODO * lint * show lint errors * fix lint * lint * lint * lint * run go generate on PR to ensure mocks are kept aligned with interfaces * fix tests * lint * go generate * fix race * return concrete type instead of interface * enforce copyright is up-to-date in PRs * naming | 1 year ago |
| update readme initial (#6523) * update readme initial * add thumbnail * Add intro, features * add logo * Add Logo * update logo * Delete .images/core-2-logo.png * update logo 2 * Delete .images/core-2-logo.png * Add files via upload * Update content, pair down unused sections * change thumbail location * Update thumbnail location * Delete docs-gb/images/Core-intro-thumbnail.png * Add docs categories * Update formatting * Fix CTA wording | 1 year ago |
| fix(docs): changed the name from core v2 to Core 2 (#5963) * changes the name from seldon core v2 to Seldon Core 2 * Update LICENSE Update after Tanu confirms from JP * Update license.txt update after Tanu confirms * Update license.txt update after Tanu confirms * Update license.txt update after Tanu confirms * Update license.txt update after Tanu confirms * Update license.txt update after Tanu confirms * Update license.txt reverted to the old file * revert the license files --------- Co-authored-by: Rakavitha Kodhandapani <seldon@SELIN002.local> | 2 years ago |
| Update CONTRIBUTING.md to add notice on licensing changes and links to previous contributions (#5555) | 2 years ago |