跳转到内容
Odysseus/docs

SRE 自动化

用途

Section titled “用途”

SRE 自动化是 SRE 编排器的控制台——SRE 编排器概念页面 描述的服务，它将警报或平台事件转化为附有建议修复方案的事件。屏幕顶部显示编排器是否启用、其当前自动化模式以及生命周期总计（引发的事件数、过去 24 小时内解决的数量、平均解决时间）；下方“近期事件”列出每个事件，可按状态和严重级别筛选，每个事件包含其警报、来源和修复计划、任务指南 逐步引导的“批准”和“拒绝”操作，以及针对无需修复审查的事件的“忽略”操作。一个单独的设置面板包含编排器将自行执行和不执行的所有内容：安全防护栏（自动修复允许的风险级别和严重级别、黑名单服务和警报、危险命令模式、每个事件的命令限制和冷却时间）、限制审查事件的模型允许执行的操作的分析设置（其系统提示、工具列表、轮次和会话超时、收集的上下文量），以及事件的 Slack 和电子邮件通知路由。

如何到达此处

Section titled “如何到达此处”

侧边栏 操作 → SRE 自动化。

外观

Section titled “外观”
Odysseus 仪表板中的SRE 自动化界面。

显示内容

Section titled “显示内容”

/sre (SREAutomation)

Section titled “/sre (SREAutomation)”
  • infrastructure:sREAutomation.0-unlimited-bounded-only-by — “0 = 无限制（仅受会话超时限制）”
  • infrastructure:sREAutomation.actions — “操作”
  • infrastructure:sREAutomation.advanced-safety-settings — “高级安全设置”
  • infrastructure:sREAutomation.alert — “告警”
  • infrastructure:sREAutomation.alert-fired — “告警已触发”
  • infrastructure:sREAutomation.alert-severities-that-can-be — “可自动修复而无需审批的告警严重级别”
  • infrastructure:sREAutomation.all-severities — “所有严重级别”
  • infrastructure:sREAutomation.all-statuses — “所有状态”
  • infrastructure:sREAutomation.allowed-risk-levels-for-auto — “自动执行允许的风险级别”
  • infrastructure:sREAutomation.allowed-severities-for-auto-remediation — “自动修复允许的严重级别”
  • infrastructure:sREAutomation.allowed-tools — “允许的工具”
  • infrastructure:sREAutomation.approve — “批准”
  • infrastructure:sREAutomation.approve-execute — “批准并执行”
  • infrastructure:sREAutomation.approving — “正在批准…”
  • infrastructure:sREAutomation.automation-mode — “自动化模式”
  • infrastructure:sREAutomation.avg-resolution-time — “平均解决时间”
  • infrastructure:sREAutomation.awaiting-approval — “等待批准”
  • infrastructure:sREAutomation.bash-read-grep-glob-webfetch — “Bash, Read, Grep, Glob, WebFetch, Task, Skill”
  • infrastructure:sREAutomation.blacklisted-alerts-comma-separated — “黑名单告警（逗号分隔）”
  • infrastructure:sREAutomation.blacklisted-services — “黑名单服务”
  • infrastructure:sREAutomation.cancel — “取消”
  • infrastructure:sREAutomation.channel — “@channel”
  • infrastructure:sREAutomation.claude-analysis-settings — “Claude 分析设置”
  • infrastructure:sREAutomation.close — “关闭”
  • infrastructure:sREAutomation.comma-separated-severities-to-notify — “用于通知的逗号分隔严重级别”
  • infrastructure:sREAutomation.command-risk-levels-that-can — “可无需审批执行的命令风险级别”
  • infrastructure:sREAutomation.command-timeout-seconds — “命令超时（秒）”
  • infrastructure:sREAutomation.commands — “命令”
  • infrastructure:sREAutomation.commands-executed — “已执行命令”
  • infrastructure:sREAutomation.container — “容器”
  • infrastructure:sREAutomation.context-gathering — “上下文收集”
  • infrastructure:sREAutomation.cooldown-seconds — “冷却时间（秒）”
  • infrastructure:sREAutomation.created — “已创建”
  • infrastructure:sREAutomation.critical — “严重”
  • infrastructure:sREAutomation.critical-alerts — “#critical-alerts”
  • infrastructure:sREAutomation.critical-channel — “严重通道”
  • infrastructure:sREAutomation.critical-warning — “critical, warning”
  • infrastructure:sREAutomation.custom-prompt-active-click-reset — “自定义提示已激活。点击“重置为默认”以恢复原始设置。”
  • infrastructure:sREAutomation.dangerous-patterns — “危险模式”
  • infrastructure:sREAutomation.days-to-keep-old-incidents — “保留旧事件的天数”
  • infrastructure:sREAutomation.default-channel — “默认渠道”
  • infrastructure:sREAutomation.description — “描述”
  • infrastructure:sREAutomation.dismiss — “关闭”
  • infrastructure:sREAutomation.dismiss-all — “全部关闭”
  • infrastructure:sREAutomation.edit — “编辑”
  • infrastructure:sREAutomation.email-routing — “邮件路由”
  • infrastructure:sREAutomation.email-subject-prefix — “邮件主题前缀”
  • infrastructure:sREAutomation.enable-or-disable-automated-incident — “启用或禁用自动化事件响应”
  • infrastructure:sREAutomation.enable-sre-email — “启用 SRE 邮件”
  • infrastructure:sREAutomation.enable-sre-slack — “启用 SRE Slack”
  • infrastructure:sREAutomation.enabled — “已启用”
  • infrastructure:sREAutomation.error — “错误：”
  • infrastructure:sREAutomation.escalated — “已升级”
  • infrastructure:sREAutomation.escalation-channel — “升级渠道”
  • infrastructure:sREAutomation.event-triggers — “事件触发器”
  • infrastructure:sREAutomation.explain-why-this-incident-is — “解释拒绝此事件的原因…”
  • infrastructure:sREAutomation.failed — “失败”
  • infrastructure:sREAutomation.general-settings — “通用设置”
  • infrastructure:sREAutomation.id — “ID：”
  • infrastructure:sREAutomation.in-progress — “进行中”
  • infrastructure:sREAutomation.incident-retention-days — “事件保留天数”
  • infrastructure:sREAutomation.include-commands — “包含命令”
  • infrastructure:sREAutomation.include-docker-stats — “包含 Docker 统计信息”
  • infrastructure:sREAutomation.include-prometheus-metrics — “包含 Prometheus 指标”
  • infrastructure:sREAutomation.include-root-cause-analysis — “包含根本原因分析”
  • infrastructure:sREAutomation.include-trace-analysis — “包含追踪分析”
  • infrastructure:sREAutomation.info — “信息”
  • infrastructure:sREAutomation.last-24-hours — “最近 24 小时”
  • infrastructure:sREAutomation.live — “实时”
  • infrastructure:sREAutomation.loading — “加载中…”
  • infrastructure:sREAutomation.log-lines-to-gather — “要收集的日志行数”
  • infrastructure:sREAutomation.max-commands-per-incident — “每个事件的最大命令数”
  • infrastructure:sREAutomation.max-time-for-each-remediation — “每个修复命令的最大时间（例如，docker restart）”
  • infrastructure:sREAutomation.max-tokens — “最大令牌数”
  • infrastructure:sREAutomation.max-turns-hard-limit — “最大轮次（硬限制）”
  • infrastructure:sREAutomation.maximum-tokens-for-claude-response — “Claude 响应的最大令牌数”
  • infrastructure:sREAutomation.message-customization — “消息自定义”
  • infrastructure:sREAutomation.no — “否”
  • infrastructure:sREAutomation.no-incidents-found — “未发现事件”
  • infrastructure:sREAutomation.no-incidents-match-the-current — “没有事件匹配当前筛选条件”
  • infrastructure:sREAutomation.no-incidents-yet — “尚无事件”
  • infrastructure:sREAutomation.no-risk-levels — “否”
  • infrastructure:sREAutomation.none-executed — “无已执行”
  • infrastructure:sREAutomation.observe-only-no-auto-execution — “仅观察 - 不自动执行”
  • infrastructure:sREAutomation.oncall — “#oncall”
  • infrastructure:sREAutomation.one-pattern-per-line-commands — “每行一个模式 - 匹配这些模式的命令将被阻止”
  • infrastructure:sREAutomation.open — “打开”
  • infrastructure:sREAutomation.polling — “轮询”
  • infrastructure:sREAutomation.postgresqldown-dataloss — “PostgreSQLDown, DataLoss”
  • infrastructure:sREAutomation.quick-dismiss — “快速关闭”
  • infrastructure:sREAutomation.reason-optional — “原因（可选）”
  • infrastructure:sREAutomation.received-by-sre — “已由 SRE 接收”
  • infrastructure:sREAutomation.recent-incidents — “近期事件”
  • infrastructure:sREAutomation.refresh — “刷新”
  • infrastructure:sREAutomation.reject — “拒绝”
  • infrastructure:sREAutomation.reject-incident — “拒绝事件”
  • infrastructure:sREAutomation.reject-incident-2 — “拒绝事件：”
  • infrastructure:sREAutomation.reject-with-reason — “带原因拒绝”
  • infrastructure:sREAutomation.rejecting — “正在拒绝…”
  • infrastructure:sREAutomation.remediation — “修复”
  • infrastructure:sREAutomation.remediation-plan — “修复计划”
  • infrastructure:sREAutomation.reset-to-default — “重置为默认”
  • infrastructure:sREAutomation.resolution-type — “解决类型”
  • infrastructure:sREAutomation.resolved — “已解决”
  • infrastructure:sREAutomation.risk-level — “风险级别”
  • infrastructure:sREAutomation.rm-rf-dd-if-drop — “rm -rf /\ndd if=\nDROP DATABASE”
  • infrastructure:sREAutomation.root-cause-analysis — “根本原因分析”
  • infrastructure:sREAutomation.safety-guardrails — “安全防护栏”
  • infrastructure:sREAutomation.save-configuration — “保存配置”
  • infrastructure:sREAutomation.saving — “保存中…”
  • infrastructure:sREAutomation.select-service-to-blacklist — “选择要加入黑名单的服务…”
  • infrastructure:sREAutomation.send-test-email — “发送测试邮件”
  • infrastructure:sREAutomation.send-test-message — “发送测试消息”
  • infrastructure:sREAutomation.sending — “发送中…”
  • infrastructure:sREAutomation.session-timeout-seconds — “会话超时（秒）”
  • infrastructure:sREAutomation.settings-notifications — “设置 → 通知”
  • infrastructure:sREAutomation.severity — “严重性”
  • infrastructure:sREAutomation.severity-filter — “严重性过滤器”
  • infrastructure:sREAutomation.slack-mention-on-critical — “严重时 Slack 提及”
  • infrastructure:sREAutomation.slack-routing — “Slack 路由”
  • infrastructure:sREAutomation.smtp-and-slack-credentials-are — “SMTP 和 Slack 凭据在以下位置配置”
  • infrastructure:sREAutomation.source — “来源”
  • infrastructure:sREAutomation.sre — “[SRE]”
  • infrastructure:sREAutomation.sre-alerts — “#sre-alerts”
  • infrastructure:sREAutomation.sre-automation — “SRE 自动化”
  • infrastructure:sREAutomation.sre-automation-enabled — “SRE 自动化已启用”
  • infrastructure:sREAutomation.sre-notification-routing — “SRE 通知路由”
  • infrastructure:sREAutomation.sre-orchestrator-status — “SRE 编排器状态”
  • infrastructure:sREAutomation.sre-will-not-take-automated — “SRE 将不会对这些服务执行自动操作”
  • infrastructure:sREAutomation.status — “状态”
  • infrastructure:sREAutomation.system-prompt — “系统提示”
  • infrastructure:sREAutomation.this-section-configures-sre-specific — “。此部分配置 SRE 特定的路由规则。”
  • infrastructure:sREAutomation.time-allowed-per-analysis-turn — “每次分析轮次允许的时间（默认：900 秒 = 15 分钟）”
  • infrastructure:sREAutomation.tools-claude-can-use-comma — “Claude 可使用的工具（逗号分隔）”
  • infrastructure:sREAutomation.total-incidents — “总事件数”
  • infrastructure:sREAutomation.total-time-for-multi-turn — “多轮分析总时长（默认：9000秒 = 2.5小时）”
  • infrastructure:sREAutomation.trace-window-minutes — “跟踪窗口（分钟）”
  • infrastructure:sREAutomation.turn-timeout-seconds — “轮次超时（秒）”
  • infrastructure:sREAutomation.unknown — “未知”
  • infrastructure:sREAutomation.use-system-default-email — “使用系统默认电子邮件”
  • infrastructure:sREAutomation.using-default-sre-prompt-click — “正在使用默认 SRE 提示词。点击“查看并编辑默认”进行自定义。”
  • infrastructure:sREAutomation.using-default-sre-prompt-click-2 — “使用默认 SRE 提示。点击“查看并编辑默认设置”以查看和修改它。”
  • infrastructure:sREAutomation.view — “查看”
  • infrastructure:sREAutomation.view-edit-default — “查看并编辑默认设置”
  • infrastructure:sREAutomation.warning — “警告”
  • infrastructure:sREAutomation.yes — “是”
  • infrastructure:sreHealthState.degraded — “正在运行，但未完全正常工作——已记录事件，请参见下文”
  • infrastructure:sreHealthState.online — “在线且运行正常”
  • infrastructure:sreHealthState.unavailable — “不可用”

可执行操作

Section titled “可执行操作”

/sre

Section titled “/sre”
  • 需要 sre:approve。请参阅角色参考]以了解哪些角色拥有此权限。
  • 需要 sre:configure。请参阅角色参考]以了解哪些角色拥有此权限。

下一步

Section titled “下一步”

关于事件是什么、如何归属到租户以及编排器将自动执行和不执行的操作，请参阅 SRE 编排器概念页面。批准排队的修复方案在 任务指南 中逐步引导。此屏幕操作所需的 sre:approve 和 sre:configure 权限以及拥有这些权限的角色在 角色参考 中；如果批准或配置更改被拒绝，拒绝参考 记录了拒绝的字段、值和可接受的形式。