Release Notes

ks2024.1-rc1-21

New Features

  • config creating and updating operations do not require the whole config body. All config attributes are optional, except “name”. Optional attributes get default values if they are not provided.

  • all recommendations for specific optimization can be applied. To do so you need to provide “optimization_id” in body of HTTP POST request to “/migrations”. Response format for this request was changed. Now it always responses with list of migrations even if there is one migration.

  • added option “sql_debug” in config group “database” which logs all SQL operations of DRS

  • hypervisor/VM are excluded from optimization if there are no metrics for them. Error log message is written.

  • attribute “metrics” of optimization is filled with raw metrics. Before this attribute was always empty.

Upgrade Notes

  • Response format of HTTP POST request to “/migrations” was changed. Now it always responds with list of migrations even if there is one migration.

  • now creating and updating operations for “job” return the whole “job” body from DB

Bug Fixes

  • fixed optimization crash if the feature “excluding from DRS” is enabled and VMs remain after the check of the feature.

  • fixed the bug when all DB configuration options (except DB URI) were not applied from configuration file and were applied default values.

  • fixed inaccurate data for all DRS objects related with users and timestamps, e.g. attributes “created_user”, “updated_user”, “created_at”, “updated_at”, “started_at”, “ended_at”, etc

  • removed legacy options from configuration file. These options were not applied in DRS, but were disrupting for user.

  • fixed sorting and filtering for all “GET” and “LIST” operations.

  • now creating and updating API methods for “job” return the whole “job” body from DB

Other Notes

  • added tox in Gitlab CI

v1.4.2

New Features

  • the configuration option “after_migrations_pause” has been added, which defines the time period during which new optimizations are not started if migrations were carried out in the previous optimization. This amount of time is necessary so that invalid metrics are not collected immediately after the migration of virtual machines.

  • added support for authorization and SSL verification for alert sending to Prometheus Alert Manager. New config options “enable_prometheus_alert_manager_auth”, “prometheus_alert_manager_user”, “prometheus_alert_manager_password” in section “alerting” were added for authorization. Existing config option “ssl_verify” in section “DEFAULT” is used for SSL verification

  • added ability to update the name and period of the job

  • hypervisor status field has been added in hypervisor scores in optimizations

  • added the last migration status field to the recommendation

  • added details of the recommendation to the migration body in the API

Known Issues

  • impossible to disable the feature of “excluding” and its types from DRS if it was enabled once. Workaround is to re-create configuration.

Upgrade Notes

  • config option “db_uri” moved from section “api” to “database” in config file

  • command-line argument “mode” was removed

  • upgrade requires database schema update

Bug Fixes

  • the feature of exclusion from drs and its types now take values from the job config, not from the config file

  • fixed the optimization crash error if the job was previously updated with authorization enabled

  • fixed an alert crash to Prometheus Alert Manager

  • fixed the operation of the ServerGroupAffinityFilter filter

  • fixed the operation of the ServerGroupAntiAffinityFilter filter

  • the values of CPU weight, RAM weight, metric collection time range, hypervisor overload threshold are now taken from the job config, and not from the config file