KIT Tier 1 AUDITOR-APEL v0.6.3 configuration

README.md

KIT Tier 1 AUDITOR-APEL v0.6.3 configuration

This is a simple configuration to have AUDITOR act as an APEL client for HTCondor. In this setup the entire AUDITOR+APEL chain runs a separate instance directly on each CE. You also need to configure the AUDITOR core and its PostgreSQL database.

If you want to run the toolchain from a different host (say the Condor Negotiator), adjust the schedd_names to cover all the condor submit nodes / CEs. If you want to support subsites, adjust the various site entries in both configurations.

apel-plugin.yaml

!Config
plugin:
  log_level: TRACE
  time_json_path: "time.json"
  report_interval: 21600
  message_type: summaries

site:
  publish_since: "2020-01-01 12:00:00+00:00"
  sites_to_report: {"FZK-LCG2":["FZK-LCG2"]}

auditor:
  ip: ...
  port: ...
  timeout: 14400
  site_meta_field: "site"

summary_fields:
  mandatory:
    NormalisedWallDuration: !NormalisedWallDurationField
      score:
        name: "HEPSPEC"
        component_name: "Cores"
    CpuDuration: !ComponentField
      name: "CPUTime"
    NormalisedCpuDuration: !NormalisedField
      base_value: !ComponentField
        name: "CPUTime"
      score:
        name: "HEPSPEC"
        component_name: "Cores"
  optional:
    GlobalUserName: !MetaField
      name: "subject"
    VO: !MetaField
      name: "voms"
      regex: '(?<=%2F).*?(?=%2F|$)'
    VOGroup: !MetaField
      name: "voms"
      regex: '(?<=%2F).*?(?=%2FRole|$)'
    VORole: !MetaField
      name: "voms"
      regex: '(?=Role).*?(?=%2F|$)'
    SubmitHost: !MetaField
      name: "submithost"
    Infrastructure: !ConstantField
      value: "grid"
    NodeCount: !ComponentField
      name: "MaxHosts"
    Processors: !ComponentField
      name: "Cores"

individual_job_fields:
  mandatory:
    CpuDuration: !ComponentField
      name: "CPUTime"

  optional:
    GlobalUserName: !MetaField
      name: "subject"
    VO: !MetaField
      name: "voms"
      regex: '(?<=%2F).*?(?=%2F|$)'
    VOGroup: !MetaField
      name: "voms"
      regex: '(?<=%2F).*?(?=%2FRole|$)'
    VORole: !MetaField
      name: "voms"
      regex: '(?=Role).*?(?=%2F|$)'
    SubmitHost: !MetaField
      name: "submithost"
    InfrastructureType: !ConstantField
      value: "grid"
    NodeCount: !ComponentField
      name: "MaxHosts"
    Processors: !ComponentField
      name: "Cores"
    FQAN: !MetaField
      name: "voms"
    ServiceLevel: !ScoreField
      name: "HEPSPEC"
      component_name: Cores
    ServiceLevelType: !ConstantField
      value: "HEPSPEC"

authentication:
  auth_url: "https://msg.argo.grnet.gr:8443/v1/service-types/ams/hosts/msg.argo.grnet.gr:authx509"
  ams_url: "https://msg.argo.grnet.gr:443/v1/projects/accounting/topics/gLite-APEL:publish?key="
  client_cert: "/etc/grid-security/hostcert.pem"
  client_key: "/etc/grid-security/hostkey.pem"
  ca_path: "/etc/grid-security/certificates"
  verify_ca: true

auditor-apel-plugin.service

[Unit]
Description=APEL plugin for AUDITOR
Documentation=https://alu-schumacher.github.io/AUDITOR/
Requires=auditor.service

[Install]
RequiredBy=multi-user.target

[Service]
Type=simple
User=root
Group=root
WorkingDirectory=/opt/auditor
ExecStart=/opt/auditor/venv/bin/auditor-apel-publish --config /opt/auditor/configuration/apel-plugin.yaml
Restart=on-failure
RestartSec=60

auditor-htcondor-collector.service

[Unit]
Description=HTCondor collector for AUDITOR
Documentation=https://alu-schumacher.github.io/AUDITOR/
Requires=auditor.service

[Install]
RequiredBy=multi-user.target

[Service]
Type=simple
User=root
Group=root
WorkingDirectory=/opt/auditor
ExecStart=/opt/auditor/venv/bin/auditor-htcondor-collector --config /opt/auditor/configuration/htcondor-collector.yaml
Restart=on-failure
RestartSec=60

htcondor-collector.yaml

addr: ...
port: ...

state_db: htcondor_history_state.db
record_prefix: htcondor
interval: 900
schedd_names:
  - htcondor-ce-3-kit.gridka.de

job_status:
  - 3 # Removed
  - 4 # Completed

meta:
  user:
    key: Owner
  group:
    key: VoName
  submithost:
    name: htcondor-ce-3-kit.gridka.de
  voms:
    key: x509UserProxyFirstFQAN
  subject:
    key: x509UserProxySubject
  site:
    - {"name":"FZK-LCG2"}

components:
  - name: "Cores"
    key: "CpusProvisioned"
    scores:
      - name: "HEPSPEC"
        key: "MachineAttrApelSpecs0.HEPSPEC"
      - name: "HEPscore23"
        key: "MachineAttrApelSpecs0.HEPscore23"
  - name: "Memory"
    key: "MemoryProvisioned"
  - name: "CPUTime"
    key: "RemoteUserCpu+RemoteSysCpu"
  - name: "Wallclocktime"
    key: "RemoteWallClockTime"
  - name: "MinHosts"
    key: "MinHosts"
  - name: "MaxHosts"
    key: "MaxHosts"