sapcc / limes

79%

master: 80%

LAST BUILD ON BRANCH acpq-buildup-improvements

branch: acpq-buildup-improvements

CHANGE BRANCH

x

Reset

• acpq-buildup-improvements
• acpq-quota-overrides-bug
• add-cluster-level-path
• add-status-to-get-commitments-api
• add_topologies
• api_test_fix
• audit-format
• audit-library
• auditevent-projectid
• auto-approve
• autoapproval-metrics
• autoupdateable-deps
• avoid-backendquota-edit-war
• baremetal-quotas
• better-log-for-project-deletion-errors
• block-expiry-notifications-as-needed
• bump-go-api-declarations
• bump-go-bits
• bump-liquidapi
• burst-audit-trail
• burst-inconsistencies
• burst-quota
• cadf-audit-log
• cadf-audit-trail
• capacity-plugin-metrics
• cfm
• cfm-recognize-duplicate-pools
• cfm-sized-used
• cfm-update
• check-metadata-integrity
• cinder_capacity
• cinder_update
• cleanup
• cli
• cluster-config-validation-testing
• commitment-merging
• commitment-table-type-cleanup
• commitment-uuid
• commitment_notification
• conf_validitaion
• consistent-plugin-metrics-delivery
• consume-liquid-utils-from-go-bits
• consumption-mode
• continuously-applied-constraints
• custom-quota-class-set
• decouple-mock-and-plugin
• delegate_commitment_acceptance
• desired-backend-quota
• docs
• domain-specific-commitment-configuration
• domain_api
• enable-liquid-info-polling
• endpoint-opts
• error-audit-events
• example
• failed_capa_scrapes
• field-test-for-option-type-2
• fix-acpq-for-ceph
• fix-azseparated-sum-quota
• fix-crash-in-topology
• fix-dupword-lint
• fix-liquid-cinder
• fix-save-rateinfo
• fix_expiration_notification
• fix_expiring_commitments
• fix_fcd
• fix_modernize
• fix_projectID
• fix_project_api
• fix_tests
• flavors
• gh-workflows
• go-bindata
• gomm-remove-coveralls
• graceful-shutdown
• ha_reserve
• handle_teapod
• health-ping
• inconsistency-detection
• initial-domain-quota
• instance-hypervisors
• instance-ip-addresses
• introduce_liquid_version_where_missing
• ironic-capacity
• ironic_limit
• ironic_override
• lbaas_optional
• liquid-ironic-fix-topology
• liquid-ironic-limit-memusage
• liquid-manila-capacity-reporting
• liquid-nova-capacity-underflow
• liquid-nova-capacity-underreporting
• liquid-nova-fix-missing-az
• liquid-nova-more-debug-output
• liquid-nova-pooled-extra-specs
• liquid-nova-report-capacity
• liquid-nova-report-usage
• liquidapi-validation
• localize_cluster_level_timestamps
• lock-down-topology-exceptions
• mail-optional
• mail_support
• manila-capacity-plugin
• manila-no-double-counting
• manila_capacity_only_default
• manual-capacity-input
• master
• match_toplogy
• max_quota
• merge-scrape-and-scrape-rates-job
• merge_capacitors_services_config
• metrics-with-servicename-label
• metrics_manila_snapmirror
• mig-netapp-metrics
• min_expiresAt_metric
• minor-fixes
• mock-liquids
• modernize
• multiple-constraints
• neutron-liquid-support-dynamic-quotas
• neutron-liquid-support-vpnaas-fwaas
• neutron-quota-details
• new-cfm
• nova_capacity
• obfuscated-errors
• os-type
• physical-usage
• plugins
• prioritize-scraping-new-services
• project_commitment_uuid_type
• project_level_v2
• quota-seeding
• rate-topology
• ratelimits
• readme-update
• refactor
• refactor-discovery-plugins
• refactor_capacitors_config
• refactoring-remove-core-service-info
• relabel_metric
• reliable-scrape-of-plugin-metrics
• remove-legacy-nova-plugins
• remove-productname
• remove-resource-data-types
• remove_cluster_capacitor_table
• remove_liquid_service_type
• remove_liquid_service_type_2
• remove_project_level_v1
• rename_cluster_level
• renew
• renovate/external-dependencies
• renovate/github.comsapcc
• replace-max-quota-with-forbidden
• resource_topology
• resource_topology_fix
• revert-liquid-cinder-changes
• revert-revert-liquid-cinder-changes
• rollup
• save_service_info_in_db
• schwift
• server-groups
• set-quota
• share-replica-quotas
• simplify-commitment-acceptance-check
• simulate-put
• subcapacities
• subresource-AZ
• subresources
• support-network-bgpvpn-quota
• swift-account-metrics
• swift_capacity
• swift_gone
• swift_plugin
• test-fix
• trace-serviceinfo-updates
• typo
• unbreak-cinder-api
• update-easypg
• use-liquid-normalize-az
• use-postgres-for-testing
• use-postlite
• use_octavia_quota_usage
• utilize_service_info_from_db2
• versatile-resources
• version_discovery
• workaround_manila

Build # 15212906791

Build Type

Pull #724

github

Committed by majewsky

Commit Message

ACPQ: do not block base quota overcommit because of empty AZs in buildup

The addition of the second AZ in Tokyo to our config is causing
problems. There is no compute capacity, but because customers have seen
the new AZ, they tried poking it and creating instances there. These
instances will inevitably fail to spawn because there are no matching
hypervisors to place them on, but the instances are still around in
status `ERROR`. When ACPQ runs, it sees usage in the AZ, but no capacity
to go with it, and so base quota overcommit is blocked.

This is an undesirable behavior, but we don't see a good way to prevent
this situation from arising. Instances in status `ERROR` need to be
reported (because they count towards the quota) and need to be in the
correct AZ (because billing needs to associate them with commitments to
apply the correct price).

We have an existing special case to disregard AZs with zero capacity and
also zero usage for the base quota overcommit (which covers the decom
case), so this gets extended to also include AZs that _never_ had
non-zero capacity (which covers the buildup case).

Pull Request Pull Request #724: ACPQ: do not block base quota overcommit because of empty AZs in buildup

Run Details

26 of 26 new or added lines in 3 files covered. (100.0%)

6447 of 8147 relevant lines covered (79.13%)

49.49 hits per line