@oneuptime/common 11.0.2 → 11.0.3

package/Server/Services/DockerSwarmClusterOwnerRuleEngineService.ts

@@ -0,0 +1,232 @@
+import Label from "../../Models/DatabaseModels/Label";
+import DockerSwarmCluster from "../../Models/DatabaseModels/DockerSwarmCluster";
+import DockerSwarmClusterOwnerRule from "../../Models/DatabaseModels/DockerSwarmClusterOwnerRule";
+import DockerSwarmClusterOwnerUser from "../../Models/DatabaseModels/DockerSwarmClusterOwnerUser";
+import DockerSwarmClusterOwnerTeam from "../../Models/DatabaseModels/DockerSwarmClusterOwnerTeam";
+import DockerSwarmClusterOwnerRuleService from "./DockerSwarmClusterOwnerRuleService";
+import DockerSwarmClusterOwnerUserService from "./DockerSwarmClusterOwnerUserService";
+import DockerSwarmClusterOwnerTeamService from "./DockerSwarmClusterOwnerTeamService";
+import DockerSwarmClusterService from "./DockerSwarmClusterService";
+import ObjectID from "../../Types/ObjectID";
+import CaptureSpan from "../Utils/Telemetry/CaptureSpan";
+import logger, { LogAttributes } from "../Utils/Logger";
+
+class DockerSwarmClusterOwnerRuleEngineServiceClass {
+  /**
+   * Evaluates DockerSwarmClusterOwnerRule rows for the given DockerSwarm cluster and adds matched
+   * owner users / teams via DockerSwarmClusterOwnerUserService / DockerSwarmClusterOwnerTeamService. Rules
+   * with notifyOwners set notify the added owners; rules with notifyOwners off
+   * add silently.
+   */
+  @CaptureSpan()
+  public async applyRulesToDockerSwarmCluster(
+    dockerSwarmCluster: DockerSwarmCluster,
+  ): Promise<void> {
+    if (!dockerSwarmCluster.id || !dockerSwarmCluster.projectId) {
+      return;
+    }
+
+    try {
+      const rules: Array<DockerSwarmClusterOwnerRule> =
+        await DockerSwarmClusterOwnerRuleService.findBy({
+          query: {
+            projectId: dockerSwarmCluster.projectId,
+            isEnabled: true,
+          },
+          props: { isRoot: true },
+          select: {
+            _id: true,
+            name: true,
+            notifyOwners: true,
+            dockerSwarmClusterLabels: { _id: true },
+            dockerSwarmClusterNamePattern: true,
+            dockerSwarmClusterDescriptionPattern: true,
+            ownerUsers: { _id: true },
+            ownerTeams: { _id: true },
+          },
+          limit: 100,
+          skip: 0,
+        });
+
+      if (rules.length === 0) {
+        return;
+      }
+
+      const dockerSwarmClusterWithDetails: DockerSwarmCluster | null =
+        await DockerSwarmClusterService.findOneById({
+          id: dockerSwarmCluster.id,
+          select: {
+            name: true,
+            description: true,
+            labels: { _id: true },
+          },
+          props: { isRoot: true },
+        });
+
+      if (!dockerSwarmClusterWithDetails) {
+        return;
+      }
+
+      const usersByNotify: Map<boolean, Set<string>> = new Map([
+        [true, new Set()],
+        [false, new Set()],
+      ]);
+      const teamsByNotify: Map<boolean, Set<string>> = new Map([
+        [true, new Set()],
+        [false, new Set()],
+      ]);
+
+      const matchedRules: Array<DockerSwarmClusterOwnerRule> = [];
+
+      for (const rule of rules) {
+        const matches: boolean = this.doesDockerSwarmClusterMatchRule(
+          dockerSwarmClusterWithDetails,
+          rule,
+        );
+        if (!matches) {
+          continue;
+        }
+        let ruleAddedAny: boolean = false;
+        const notify: boolean = rule.notifyOwners !== false;
+        for (const user of rule.ownerUsers || []) {
+          if (user.id) {
+            usersByNotify.get(notify)!.add(user.id.toString());
+            ruleAddedAny = true;
+          }
+        }
+        for (const team of rule.ownerTeams || []) {
+          if (team.id) {
+            teamsByNotify.get(notify)!.add(team.id.toString());
+            ruleAddedAny = true;
+          }
+        }
+        if (ruleAddedAny) {
+          matchedRules.push(rule);
+        }
+      }
+
+      if (matchedRules.length === 0) {
+        return;
+      }
+
+      for (const notify of [true, false]) {
+        const userIds: Set<string> = usersByNotify.get(notify)!;
+        const teamIds: Set<string> = teamsByNotify.get(notify)!;
+
+        for (const userId of userIds) {
+          const owner: DockerSwarmClusterOwnerUser =
+            new DockerSwarmClusterOwnerUser();
+          owner.dockerSwarmClusterId = dockerSwarmCluster.id;
+          owner.projectId = dockerSwarmCluster.projectId;
+          owner.userId = new ObjectID(userId);
+          owner.isOwnerNotified = !notify;
+          await DockerSwarmClusterOwnerUserService.create({
+            data: owner,
+            props: { isRoot: true },
+          });
+        }
+
+        for (const teamId of teamIds) {
+          const owner: DockerSwarmClusterOwnerTeam =
+            new DockerSwarmClusterOwnerTeam();
+          owner.dockerSwarmClusterId = dockerSwarmCluster.id;
+          owner.projectId = dockerSwarmCluster.projectId;
+          owner.teamId = new ObjectID(teamId);
+          owner.isOwnerNotified = !notify;
+          await DockerSwarmClusterOwnerTeamService.create({
+            data: owner,
+            props: { isRoot: true },
+          });
+        }
+      }
+
+      logger.debug(
+        `DockerSwarmClusterOwnerRuleEngine added owners to DockerSwarm cluster ${dockerSwarmCluster.id}`,
+        { projectId: dockerSwarmCluster.projectId.toString() } as LogAttributes,
+      );
+    } catch (error) {
+      logger.error(`Error applying DockerSwarm cluster owner rules: ${error}`, {
+        projectId: dockerSwarmCluster.projectId?.toString(),
+        dockerSwarmClusterId: dockerSwarmCluster.id?.toString(),
+      } as LogAttributes);
+    }
+  }
+
+  private doesDockerSwarmClusterMatchRule(
+    dockerSwarmCluster: DockerSwarmCluster,
+    rule: DockerSwarmClusterOwnerRule,
+  ): boolean {
+    if (
+      rule.dockerSwarmClusterLabels &&
+      rule.dockerSwarmClusterLabels.length > 0
+    ) {
+      if (
+        !dockerSwarmCluster.labels ||
+        dockerSwarmCluster.labels.length === 0
+      ) {
+        return false;
+      }
+      const ruleLabelIds: Array<string> = rule.dockerSwarmClusterLabels.map(
+        (l: Label) => {
+          return l.id?.toString() || "";
+        },
+      );
+      const labelIds: Array<string> = dockerSwarmCluster.labels.map(
+        (l: Label) => {
+          return l.id?.toString() || "";
+        },
+      );
+      if (
+        !ruleLabelIds.some((id: string) => {
+          return labelIds.includes(id);
+        })
+      ) {
+        return false;
+      }
+    }
+
+    if (
+      rule.dockerSwarmClusterNamePattern &&
+      (!dockerSwarmCluster.name ||
+        !this.testRegex(
+          rule.dockerSwarmClusterNamePattern,
+          dockerSwarmCluster.name,
+          rule,
+        ))
+    ) {
+      return false;
+    }
+
+    if (
+      rule.dockerSwarmClusterDescriptionPattern &&
+      (!dockerSwarmCluster.description ||
+        !this.testRegex(
+          rule.dockerSwarmClusterDescriptionPattern,
+          dockerSwarmCluster.description,
+          rule,
+        ))
+    ) {
+      return false;
+    }
+
+    return true;
+  }
+
+  private testRegex(
+    pattern: string,
+    value: string,
+    rule: DockerSwarmClusterOwnerRule,
+  ): boolean {
+    try {
+      const regex: RegExp = new RegExp(pattern, "i");
+      return regex.test(value);
+    } catch {
+      logger.warn(
+        `Invalid regex in DockerSwarm cluster owner rule ${rule.id}: ${pattern}`,
+      );
+      return false;
+    }
+  }
+}
+
+export default new DockerSwarmClusterOwnerRuleEngineServiceClass();

package/Server/Services/DockerSwarmClusterOwnerRuleService.ts

@@ -0,0 +1,14 @@
+import DatabaseService from "./DatabaseService";
+import Model from "../../Models/DatabaseModels/DockerSwarmClusterOwnerRule";
+import { IsBillingEnabled } from "../EnvironmentConfig";
+
+export class Service extends DatabaseService<Model> {
+  public constructor() {
+    super(Model);
+    if (IsBillingEnabled) {
+      this.hardDeleteItemsOlderThanInDays("createdAt", 3 * 365);
+    }
+  }
+}
+
+export default new Service();

package/Server/Services/DockerSwarmClusterOwnerTeamService.ts

@@ -0,0 +1,10 @@
+import DatabaseService from "./DatabaseService";
+import Model from "../../Models/DatabaseModels/DockerSwarmClusterOwnerTeam";
+
+export class Service extends DatabaseService<Model> {
+  public constructor() {
+    super(Model);
+  }
+}
+
+export default new Service();

package/Server/Services/DockerSwarmClusterOwnerUserService.ts

@@ -0,0 +1,10 @@
+import DatabaseService from "./DatabaseService";
+import Model from "../../Models/DatabaseModels/DockerSwarmClusterOwnerUser";
+
+export class Service extends DatabaseService<Model> {
+  public constructor() {
+    super(Model);
+  }
+}
+
+export default new Service();

package/Server/Services/DockerSwarmClusterService.ts

@@ -0,0 +1,404 @@
+import DatabaseService from "./DatabaseService";
+import DockerSwarmClusterLabelRuleEngineService from "./DockerSwarmClusterLabelRuleEngineService";
+import DockerSwarmClusterOwnerRuleEngineService from "./DockerSwarmClusterOwnerRuleEngineService";
+import Model from "../../Models/DatabaseModels/DockerSwarmCluster";
+import Label from "../../Models/DatabaseModels/Label";
+import { OnCreate } from "../Types/Database/Hooks";
+import CaptureSpan from "../Utils/Telemetry/CaptureSpan";
+import ObjectID from "../../Types/ObjectID";
+import QueryHelper from "../Types/Database/QueryHelper";
+import OneUptimeDate from "../../Types/Date";
+import LIMIT_MAX from "../../Types/Database/LimitMax";
+import GlobalCache from "../Infrastructure/GlobalCache";
+import logger, { LogAttributes } from "../Utils/Logger";
+import crypto from "crypto";
+
+const LAST_SEEN_CACHE_NAMESPACE: string = "docker-swarm-cluster-last-seen";
+const LAST_SEEN_THROTTLE_SECONDS: number = 60;
+
+const LABELS_APPLIED_CACHE_NAMESPACE: string =
+  "docker-swarm-cluster-labels-applied";
+const LABELS_APPLIED_CACHE_TTL_SECONDS: number = 60;
+
+export class Service extends DatabaseService<Model> {
+  public constructor() {
+    super(Model);
+  }
+
+  @CaptureSpan()
+  protected override async onCreateSuccess(
+    _onCreate: OnCreate<Model>,
+    createdItem: Model,
+  ): Promise<Model> {
+    /*
+     * Rules run once, on creation only — exact parity with
+     * KubernetesClusterService. Label engine first: it syncs the
+     * in-memory labels so the owner engine can match rule-added labels.
+     */
+    if (createdItem.projectId && createdItem.id) {
+      Promise.resolve()
+        .then(async () => {
+          await DockerSwarmClusterLabelRuleEngineService.applyRulesToDockerSwarmCluster(
+            createdItem,
+          );
+        })
+        .then(async () => {
+          await DockerSwarmClusterOwnerRuleEngineService.applyRulesToDockerSwarmCluster(
+            createdItem,
+          );
+        })
+        .catch((error: Error) => {
+          logger.error(
+            `Error applying dockerSwarm cluster rules in DockerSwarmClusterService.onCreateSuccess: ${error}`,
+            {
+              projectId: createdItem.projectId?.toString(),
+              dockerSwarmClusterId: createdItem.id?.toString(),
+            } as LogAttributes,
+          );
+        });
+    }
+    return createdItem;
+  }
+
+  @CaptureSpan()
+  public async findOrCreateByName(data: {
+    projectId: ObjectID;
+    name: string;
+  }): Promise<Model> {
+    /*
+     * A DockerSwarm cluster is keyed by the `docker.swarm.cluster.name` OTel resource
+     * attribute, which the user configures on the agent. Look it up
+     * case-insensitively: the unique guard (checkUniqueColumnBy ->
+     * findWithSameText) compares case-insensitively, so a case-sensitive
+     * lookup would miss an existing row that differs only by case, then
+     * fail to create it — wedging ingest for that cluster. Unlike
+     * DockerHost.hostIdentifier (host.name casing is unstable on Windows)
+     * the configured cluster name's casing is stable, so we preserve the
+     * user's casing on create instead of canonicalizing to lowercase.
+     */
+    const name: string = data.name.trim();
+
+    const existingCluster: Model | null = await this.findOneBy({
+      query: {
+        projectId: data.projectId,
+        name: QueryHelper.findWithSameText(name),
+      },
+      select: {
+        _id: true,
+        projectId: true,
+        name: true,
+      },
+      props: {
+        isRoot: true,
+      },
+    });
+
+    if (existingCluster) {
+      return existingCluster;
+    }
+
+    try {
+      // Create new cluster
+      const newCluster: Model = new Model();
+      newCluster.projectId = data.projectId;
+      newCluster.name = name;
+      newCluster.otelCollectorStatus = "connected";
+      newCluster.lastSeenAt = OneUptimeDate.getCurrentDate();
+
+      const createdCluster: Model = await this.create({
+        data: newCluster,
+        props: {
+          isRoot: true,
+        },
+      });
+
+      return createdCluster;
+    } catch {
+      /*
+       * Either two ingest workers raced to create the same cluster, or a
+       * cluster with this name in a different case already existed and the
+       * unique guard rejected the insert. Re-resolve case-insensitively so
+       * the caller still gets the existing row instead of throwing.
+       */
+      const reFetchedCluster: Model | null = await this.findOneBy({
+        query: {
+          projectId: data.projectId,
+          name: QueryHelper.findWithSameText(name),
+        },
+        select: {
+          _id: true,
+          projectId: true,
+          name: true,
+        },
+        props: {
+          isRoot: true,
+        },
+      });
+
+      if (reFetchedCluster) {
+        return reFetchedCluster;
+      }
+
+      throw new Error("Failed to create or find DockerSwarm cluster: " + name);
+    }
+  }
+
+  /*
+   * Refresh lastSeenAt / connection status and (optionally) the
+   * snapshot columns the list page renders. Count columns ride this
+   * extras path with COALESCE-per-column semantics: a key that is
+   * undefined is simply not written, so a partial batch (one that
+   * lacked the matching *_info series) never zeroes a count. The
+   * 60-second extras fingerprint cache is the write throttle — the
+   * steady state (identical snapshot every scrape) costs one Redis
+   * read per batch and at most one Postgres UPDATE per minute.
+   *
+   * Two callers share this throttle with DISJOINT extras shapes: the
+   * metrics snapshot flush (pveVersion + counts, every batch) and the
+   * fenced autoDiscoverDockerSwarmCluster maintenance path (agentVersion
+   * only — and usually an all-null fingerprint, since the shipped
+   * agent config does not stamp oneuptime.agent.version). The single
+   * fingerprint covers the whole extras object, so each alternation
+   * between the two shapes busts the throttle: at most one extra
+   * Postgres UPDATE per maintenance-fence window (~5 min), which is
+   * accepted. Do NOT key the cache per-caller — that would let two
+   * callers each refresh lastSeenAt under their own throttle and is
+   * not worth the complexity for one UPDATE per 5 minutes.
+   */
+  @CaptureSpan()
+  public async updateLastSeen(
+    clusterId: ObjectID,
+    extra?: {
+      dockerVersion?: string | undefined;
+      swarmId?: string | undefined;
+      agentVersion?: string | undefined;
+      nodeCount?: number | undefined;
+      readyNodeCount?: number | undefined;
+      managerNodeCount?: number | undefined;
+      serviceCount?: number | undefined;
+      taskCount?: number | undefined;
+      runningTaskCount?: number | undefined;
+      stackCount?: number | undefined;
+      networkCount?: number | undefined;
+    },
+  ): Promise<void> {
+    const cacheKey: string = clusterId.toString();
+    const extrasFingerprint: string = crypto
+      .createHash("sha1")
+      .update(
+        JSON.stringify({
+          dockerVersion: extra?.dockerVersion ?? null,
+          swarmId: extra?.swarmId ?? null,
+          agentVersion: extra?.agentVersion ?? null,
+          nodeCount: extra?.nodeCount ?? null,
+          readyNodeCount: extra?.readyNodeCount ?? null,
+          managerNodeCount: extra?.managerNodeCount ?? null,
+          serviceCount: extra?.serviceCount ?? null,
+          taskCount: extra?.taskCount ?? null,
+          runningTaskCount: extra?.runningTaskCount ?? null,
+          stackCount: extra?.stackCount ?? null,
+          networkCount: extra?.networkCount ?? null,
+        }),
+      )
+      .digest("hex");
+
+    const cached: string | null = await GlobalCache.getString(
+      LAST_SEEN_CACHE_NAMESPACE,
+      cacheKey,
+    );
+
+    if (cached === extrasFingerprint) {
+      return; // same data was written recently
+    }
+
+    await GlobalCache.setString(
+      LAST_SEEN_CACHE_NAMESPACE,
+      cacheKey,
+      extrasFingerprint,
+      { expiresInSeconds: LAST_SEEN_THROTTLE_SECONDS },
+    );
+
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    const data: any = {
+      lastSeenAt: OneUptimeDate.getCurrentDate(),
+      otelCollectorStatus: "connected",
+    };
+
+    if (extra?.dockerVersion) {
+      data.dockerVersion = extra.dockerVersion;
+    }
+    if (extra?.swarmId) {
+      data.swarmId = extra.swarmId;
+    }
+    if (extra?.agentVersion) {
+      data.agentVersion = extra.agentVersion;
+    }
+    // Counts: 0 is a legitimate value — gate on undefined, not falsiness.
+    if (extra?.nodeCount !== undefined) {
+      data.nodeCount = extra.nodeCount;
+    }
+    if (extra?.readyNodeCount !== undefined) {
+      data.readyNodeCount = extra.readyNodeCount;
+    }
+    if (extra?.managerNodeCount !== undefined) {
+      data.managerNodeCount = extra.managerNodeCount;
+    }
+    if (extra?.serviceCount !== undefined) {
+      data.serviceCount = extra.serviceCount;
+    }
+    if (extra?.taskCount !== undefined) {
+      data.taskCount = extra.taskCount;
+    }
+    if (extra?.runningTaskCount !== undefined) {
+      data.runningTaskCount = extra.runningTaskCount;
+    }
+    if (extra?.stackCount !== undefined) {
+      data.stackCount = extra.stackCount;
+    }
+    if (extra?.networkCount !== undefined) {
+      data.networkCount = extra.networkCount;
+    }
+
+    await this.updateOneById({
+      id: clusterId,
+      data: data,
+      props: {
+        isRoot: true,
+      },
+    });
+  }
+
+  /**
+   * Additively attach labels to a DockerSwarm cluster. Existing labels are
+   * never removed — manual labels set via the UI survive ingest. The
+   * set of labelIds passed in is fingerprinted and cached for 60s so
+   * the common case (steady-state collector pushing the same label
+   * set every batch) costs one in-memory lookup, not a join-table
+   * scan.
+   */
+  @CaptureSpan()
+  public async attachLabels(data: {
+    dockerSwarmClusterId: ObjectID;
+    labelIds: Array<ObjectID>;
+  }): Promise<void> {
+    if (!data.labelIds || data.labelIds.length === 0) {
+      return;
+    }
+
+    const cacheKey: string = data.dockerSwarmClusterId.toString();
+    const fingerprint: string = fingerprintLabelIds(data.labelIds);
+    const cached: string | null = await GlobalCache.getString(
+      LABELS_APPLIED_CACHE_NAMESPACE,
+      cacheKey,
+    );
+    if (cached === fingerprint) {
+      return;
+    }
+
+    try {
+      const dockerSwarmClusterIdStr: string =
+        data.dockerSwarmClusterId.toString();
+      const existingLabels: Array<Label> = await this.getRepository()
+        .createQueryBuilder()
+        .relation(Model, "labels")
+        .of(dockerSwarmClusterIdStr)
+        .loadMany();
+
+      const existingIds: Set<string> = new Set();
+      for (const lbl of existingLabels) {
+        const idStr: string | undefined = lbl._id?.toString();
+        if (idStr) {
+          existingIds.add(idStr);
+        }
+      }
+
+      const toAddIds: Array<string> = [];
+      const seen: Set<string> = new Set();
+      for (const id of data.labelIds) {
+        const idStr: string = id.toString();
+        if (existingIds.has(idStr) || seen.has(idStr)) {
+          continue;
+        }
+        seen.add(idStr);
+        toAddIds.push(idStr);
+      }
+
+      if (toAddIds.length > 0) {
+        await this.getRepository()
+          .createQueryBuilder()
+          .relation(Model, "labels")
+          .of(dockerSwarmClusterIdStr)
+          .add(toAddIds);
+      }
+
+      await GlobalCache.setString(
+        LABELS_APPLIED_CACHE_NAMESPACE,
+        cacheKey,
+        fingerprint,
+        { expiresInSeconds: LABELS_APPLIED_CACHE_TTL_SECONDS },
+      );
+    } catch (err) {
+      logger.warn(
+        `DockerSwarmClusterService.attachLabels failed for dockerSwarm cluster ${data.dockerSwarmClusterId.toString()}: ${
+          err instanceof Error ? err.message : String(err)
+        }`,
+      );
+    }
+  }
+
+  @CaptureSpan()
+  public async markDisconnectedClusters(): Promise<void> {
+    /*
+     * Threshold must stay well above the 5-minute OTel ingest
+     * maintenance fence (MAINTENANCE_FENCE_TTL_SECONDS in
+     * OtelIngestBaseService) — lastSeenAt is legitimately up to
+     * ~5 minutes stale during continuous telemetry, so a threshold
+     * equal to the fence TTL flaps healthy resources. 15 minutes
+     * gives 3x headroom.
+     */
+    const fifteenMinutesAgo: Date = OneUptimeDate.addRemoveMinutes(
+      OneUptimeDate.getCurrentDate(),
+      -15,
+    );
+
+    const connectedClusters: Array<Model> = await this.findBy({
+      query: {
+        otelCollectorStatus: "connected",
+        lastSeenAt: QueryHelper.lessThan(fifteenMinutesAgo),
+      },
+      select: {
+        _id: true,
+      },
+      limit: LIMIT_MAX,
+      skip: 0,
+      props: {
+        isRoot: true,
+      },
+    });
+
+    for (const cluster of connectedClusters) {
+      if (cluster._id) {
+        await this.updateOneById({
+          id: new ObjectID(cluster._id.toString()),
+          data: {
+            otelCollectorStatus: "disconnected",
+          },
+          props: {
+            isRoot: true,
+          },
+        });
+      }
+    }
+  }
+}
+
+function fingerprintLabelIds(labelIds: Array<ObjectID>): string {
+  const sorted: Array<string> = labelIds
+    .map((id: ObjectID) => {
+      return id.toString();
+    })
+    .sort();
+  return crypto.createHash("sha1").update(sorted.join(",")).digest("hex");
+}
+
+export default new Service();