omnigateway 0.8.0 → 0.8.1

package/README.md

@@ -516,6 +516,15 @@ bun apps/gateway/src/index.ts >> /var/log/omni.log 2>&1
 `omni start` does both for the gateway it supervises, and under systemd the journal needs no
 setup.
 
+In a fleet, capture is per process. Each replica can capture its own — tee its stdout to a file
+inside the container and point `OMNI_LOG_FILE` at the same path — and the Console screen then
+merges every process's tail and lets you pick one. That is worth having for an incident on a
+running pod, and it is not a log stack: the file dies with the container, nothing rotates it, and
+the screen reads one process at a time. Ship stdout to a collector for anything beyond that —
+Elasticsearch and Kibana, Loki and Grafana, or whatever already reads your containers — where the
+lines outlive the process that wrote them and can be searched across all of them at once. The
+Console screen says so itself when it finds a fleet capturing nothing.
+
 Everything else lives in the database rather than the environment, so it can be
 changed without a restart: the six routing weights, `maxAttempts`,
 `requestDeadlineMs`, the circuit breaker's `breakerThreshold` and
@@ -641,9 +650,12 @@ format, structural bounds, and masking rules.
 ## Snapshots and restore
 
 The console's Database screen reports what this installation occupies — the
-database file, its write-ahead log, the captured-body tree, and the free pages a
-compaction would give back — and takes snapshots. `omni db stats` prints the same
-figures.
+database file, its write-ahead log, the captured-body tree, the free pages a
+compaction would give back, and every table by size — and takes snapshots.
+`omni db stats` prints the same figures. On Postgres both show the server's own
+size, the `request_bodies` table and the per-table listing instead; there is no
+file, so nothing here compacts, snapshots or restores it — `pg_dump` is the
+backup.
 
 **What a snapshot is.** One self-contained SQLite file, written into a
 `snapshots/` directory beside the database. The write-ahead log is folded in, so
@@ -737,7 +749,9 @@ and the workflow commits that version into `k8s/kustomization.yaml` on `main`, w
 CD syncs. Rolling back is editing `newTag` by hand.
 
 Plugins in a fleet are baked into the image so every replica holds the same
-set: `COPY plugins/ /data/plugins/` in a derived Dockerfile.
+set: `COPY plugins/ /data/plugins/` in a derived Dockerfile. The image is
+public, so no pull secret is configured; a private fork adds
+`imagePullSecrets` to the Deployment.
 
 Give the container a restart policy — `--restart unless-stopped` — if you want a
 restart request to bring it back. A container cannot read its own policy, so

package/bin/omni.js

@@ -16281,6 +16281,23 @@ function createMaintenanceRepo(sql, nodeId) {
         schemaVersion: row?.version ?? 0
       };
     },
+    async tables() {
+      const rows = await sql.unsafe(`SELECT c.relname AS name,
+                pg_total_relation_size(c.oid) AS bytes,
+                GREATEST(c.reltuples, 0)::bigint AS rows,
+                COALESCE(s.n_dead_tup, 0) AS dead_rows
+           FROM pg_class c
+           JOIN pg_namespace n ON n.oid = c.relnamespace
+           LEFT JOIN pg_stat_user_tables s ON s.relid = c.oid
+          WHERE c.relkind = 'r' AND n.nspname = current_schema()
+          ORDER BY bytes DESC, name`);
+      return rows.map((row) => ({
+        name: row.name,
+        bytes: num(row.bytes),
+        rows: num(row.rows),
+        deadRows: num(row.dead_rows)
+      }));
+    },
     vacuum: async () => unsupported(),
     snapshotTo: async () => unsupported(),
     inspect: async () => unsupported()
@@ -17219,6 +17236,7 @@ async function createPostgresStore(opts) {
     });
   };
   return {
+    engine: "postgres",
     databasePath: redacted(opts.url),
     credentials: createCredentialRepo(sql, opts.encryptionKey, emit2),
     config: createConfigRepo(sql, emit2),
@@ -18401,6 +18419,15 @@ function createMaintenanceRepo2(db, nodeId) {
         schemaVersion: version2?.version ?? 0
       };
     },
+    async tables() {
+      return db.query(`SELECT m.tbl_name AS name,
+                  SUM(s.pgsize) AS bytes,
+                  SUM(CASE WHEN s.pagetype = 'leaf' AND s.name = m.tbl_name THEN s.ncell ELSE 0 END) AS rows
+             FROM dbstat s JOIN sqlite_master m ON m.name = s.name
+            WHERE m.tbl_name NOT LIKE 'sqlite_%'
+            GROUP BY m.tbl_name
+            ORDER BY bytes DESC, name`).all().map((row) => ({ ...row, deadRows: null }));
+    },
     async vacuum() {
       db.run("VACUUM");
       db.run("PRAGMA wal_checkpoint(TRUNCATE)");
@@ -18872,6 +18899,7 @@ async function createStore(opts) {
     live = false;
   };
   return {
+    engine: "sqlite",
     databasePath: opts.path,
     credentials: {
       list: () => handle.credentials.list(),
@@ -18936,6 +18964,7 @@ async function createStore(opts) {
     },
     maintenance: {
       stats: () => handle.maintenance.stats(),
+      tables: () => handle.maintenance.tables(),
       heartbeat: (now) => handle.maintenance.heartbeat(now),
       nodes: (now) => handle.maintenance.nodes(now),
       vacuum: () => handle.maintenance.vacuum(),
@@ -20975,13 +21004,13 @@ function encodeSystemTurn(content, cloak) {
   return content.map((block) => encodeBlock(block, cloak));
 }
 function systemCacheControl(req) {
-  const cacheable = req.messages.flatMap((message) => message.content.flatMap((block) => block.type === "thinking" ? [] : [{ role: message.role, block }]));
-  const markedSystemBlocks = cacheable.filter(({ role, block }) => role === "system" && cacheControlOf(block) !== undefined);
-  const final = markedSystemBlocks.at(-1);
-  const promoted = final !== undefined && final.block.type === "text" && cacheable.at(-1) === final ? cacheControlOf(final.block) : undefined;
+  const markedSystemBlocks = req.messages.flatMap((message) => message.role === "system" ? message.content.filter((block) => cacheControlOf(block) !== undefined) : []);
+  const trailing = req.messages.at(-1);
+  const marked = trailing?.role === "system" ? trailing.content.findLast((block) => cacheControlOf(block) !== undefined) : undefined;
+  const retarget = marked === undefined ? undefined : cacheControlOf(marked);
   return {
-    ...promoted === undefined ? {} : { promoted },
-    lost: markedSystemBlocks.length > (promoted === undefined ? 0 : 1)
+    ...retarget === undefined ? {} : { retarget },
+    lost: markedSystemBlocks.length > (retarget === undefined ? 0 : 1)
   };
 }
 function encodeTool(t, cloak) {
@@ -21143,14 +21172,26 @@ function toWire(req, model, opts) {
       return [{ role: m.role, content: encodeSystemTurn(m.content, cloak) }];
     }),
     max_tokens: req.maxTokens ?? 4096,
-    stream: req.stream,
-    ...systemCache.promoted === undefined ? {} : {
-      cache_control: {
-        type: systemCache.promoted.type,
-        ...systemCache.promoted.ttl === undefined ? {} : { ttl: systemCache.promoted.ttl }
+    stream: req.stream
+  };
+  if (systemCache.retarget !== undefined) {
+    const trailing = body.messages.at(-1);
+    if (isRecord(trailing) && Array.isArray(trailing.content)) {
+      for (const block of trailing.content) {
+        if (isRecord(block))
+          delete block.cache_control;
       }
     }
-  };
+    if (!hasVendorCacheControl(req)) {
+      const target = lastCacheableHistoryBlock(body.messages.slice(0, -1));
+      if (target === undefined) {
+        note2("anthropic:system-turn-cache-control-dropped");
+      } else if (target.cache_control === undefined) {
+        Object.assign(target, wireCacheControl(systemCache.retarget));
+        note2("anthropic:system-turn-cache-control-retargeted");
+      }
+    }
+  }
   if (system !== undefined && system.length > 0)
     body.system = system;
   if (req.temperature !== undefined)
@@ -25412,18 +25453,37 @@ async function vacuum(deps) {
 }
 async function getDatabaseOverview(deps) {
   const live = deps.store.databasePath;
-  const stats = await deps.store.maintenance.stats();
-  const settings = await deps.store.config.getSettings();
+  const [stats, tables, settings] = await Promise.all([
+    deps.store.maintenance.stats(),
+    deps.store.maintenance.tables(),
+    deps.store.config.getSettings()
+  ]);
+  const shared = {
+    engine: deps.store.engine,
+    location: live,
+    stats,
+    logicalBytes: stats.pageSize * stats.pageCount,
+    freePageBytes: stats.pageSize * stats.freelistCount,
+    retention: retentionOf(settings),
+    tables
+  };
+  if (deps.store.engine === "postgres") {
+    return {
+      ...shared,
+      fileBytes: 0,
+      walBytes: 0,
+      bodiesBytes: tables.find((table2) => table2.name === "request_bodies")?.bytes ?? 0,
+      freeDiskBytes: null,
+      snapshots: { count: 0, totalBytes: 0, latestAt: null }
+    };
+  }
   const snapshots = listSnapshots(deps);
   return {
-    stats,
+    ...shared,
     fileBytes: deps.fs.stat(live)?.size ?? 0,
     walBytes: deps.fs.stat(`${live}-wal`)?.size ?? 0,
     bodiesBytes: deps.fs.dirBytes(bodiesDirFor(live)),
-    logicalBytes: stats.pageSize * stats.pageCount,
-    freePageBytes: stats.pageSize * stats.freelistCount,
     freeDiskBytes: deps.fs.freeBytes(dirname2(live)),
-    retention: retentionOf(settings),
     snapshots: {
       count: snapshots.length,
       totalBytes: snapshots.reduce((sum2, snapshot) => sum2 + snapshot.sizeBytes, 0),
@@ -28447,7 +28507,13 @@ var dbStats = {
   async run(_args, env2) {
     const { ctx, writer } = env2;
     const overview = await getDatabaseOverview(await database(env2));
-    emit(ctx, writer, overview, () => fields([
+    const tables = table([{ header: "TABLE" }, { header: "SIZE", align: "right" }, { header: "ROWS", align: "right" }], overview.tables.map((t) => [t.name, formatBytes(t.bytes), String(t.rows)]));
+    const summary = overview.engine === "postgres" ? fields([
+      ["database", overview.location],
+      ["size", formatBytes(overview.logicalBytes)],
+      ["captured bodies", `${formatBytes(overview.bodiesBytes)} (request_bodies table)`],
+      ["schema version", String(overview.stats.schemaVersion)]
+    ]) : fields([
       ["database", ctx.databasePath],
       ["size", formatBytes(overview.fileBytes)],
       ["write-ahead log", formatBytes(overview.walBytes)],
@@ -28466,7 +28532,10 @@ var dbStats = {
         "free disk",
         overview.freeDiskBytes === null ? "unknown" : formatBytes(overview.freeDiskBytes)
       ]
-    ]));
+    ]);
+    emit(ctx, writer, overview, () => `${summary}
+
+${tables}`);
   }
 };
 var dbSnapshots = {
@@ -29300,7 +29369,7 @@ function createServiceDeps(input) {
 }
 
 // apps/cli/src/version.ts
-var VERSION = "0.8.0";
+var VERSION = "0.8.1";
 
 // apps/cli/src/commands/service.ts
 function gatewayArgv(root) {

package/gateway.js

@@ -21710,6 +21710,23 @@ function createMaintenanceRepo(sql, nodeId) {
         schemaVersion: row?.version ?? 0
       };
     },
+    async tables() {
+      const rows = await sql.unsafe(`SELECT c.relname AS name,
+                pg_total_relation_size(c.oid) AS bytes,
+                GREATEST(c.reltuples, 0)::bigint AS rows,
+                COALESCE(s.n_dead_tup, 0) AS dead_rows
+           FROM pg_class c
+           JOIN pg_namespace n ON n.oid = c.relnamespace
+           LEFT JOIN pg_stat_user_tables s ON s.relid = c.oid
+          WHERE c.relkind = 'r' AND n.nspname = current_schema()
+          ORDER BY bytes DESC, name`);
+      return rows.map((row) => ({
+        name: row.name,
+        bytes: num(row.bytes),
+        rows: num(row.rows),
+        deadRows: num(row.dead_rows)
+      }));
+    },
     vacuum: async () => unsupported(),
     snapshotTo: async () => unsupported(),
     inspect: async () => unsupported()
@@ -22648,6 +22665,7 @@ async function createPostgresStore(opts) {
     });
   };
   return {
+    engine: "postgres",
     databasePath: redacted(opts.url),
     credentials: createCredentialRepo(sql, opts.encryptionKey, emit),
     config: createConfigRepo(sql, emit),
@@ -23830,6 +23848,15 @@ function createMaintenanceRepo2(db, nodeId) {
         schemaVersion: version2?.version ?? 0
       };
     },
+    async tables() {
+      return db.query(`SELECT m.tbl_name AS name,
+                  SUM(s.pgsize) AS bytes,
+                  SUM(CASE WHEN s.pagetype = 'leaf' AND s.name = m.tbl_name THEN s.ncell ELSE 0 END) AS rows
+             FROM dbstat s JOIN sqlite_master m ON m.name = s.name
+            WHERE m.tbl_name NOT LIKE 'sqlite_%'
+            GROUP BY m.tbl_name
+            ORDER BY bytes DESC, name`).all().map((row) => ({ ...row, deadRows: null }));
+    },
     async vacuum() {
       db.run("VACUUM");
       db.run("PRAGMA wal_checkpoint(TRUNCATE)");
@@ -24301,6 +24328,7 @@ async function createStore(opts) {
     live = false;
   };
   return {
+    engine: "sqlite",
     databasePath: opts.path,
     credentials: {
       list: () => handle.credentials.list(),
@@ -24365,6 +24393,7 @@ async function createStore(opts) {
     },
     maintenance: {
       stats: () => handle.maintenance.stats(),
+      tables: () => handle.maintenance.tables(),
       heartbeat: (now) => handle.maintenance.heartbeat(now),
       nodes: (now) => handle.maintenance.nodes(now),
       vacuum: () => handle.maintenance.vacuum(),
@@ -26534,13 +26563,13 @@ function encodeSystemTurn(content, cloak) {
   return content.map((block) => encodeBlock(block, cloak));
 }
 function systemCacheControl(req) {
-  const cacheable = req.messages.flatMap((message) => message.content.flatMap((block) => block.type === "thinking" ? [] : [{ role: message.role, block }]));
-  const markedSystemBlocks = cacheable.filter(({ role, block }) => role === "system" && cacheControlOf(block) !== undefined);
-  const final = markedSystemBlocks.at(-1);
-  const promoted = final !== undefined && final.block.type === "text" && cacheable.at(-1) === final ? cacheControlOf(final.block) : undefined;
+  const markedSystemBlocks = req.messages.flatMap((message) => message.role === "system" ? message.content.filter((block) => cacheControlOf(block) !== undefined) : []);
+  const trailing = req.messages.at(-1);
+  const marked = trailing?.role === "system" ? trailing.content.findLast((block) => cacheControlOf(block) !== undefined) : undefined;
+  const retarget = marked === undefined ? undefined : cacheControlOf(marked);
   return {
-    ...promoted === undefined ? {} : { promoted },
-    lost: markedSystemBlocks.length > (promoted === undefined ? 0 : 1)
+    ...retarget === undefined ? {} : { retarget },
+    lost: markedSystemBlocks.length > (retarget === undefined ? 0 : 1)
   };
 }
 function encodeTool(t, cloak) {
@@ -26702,14 +26731,26 @@ function toWire(req, model, opts) {
       return [{ role: m.role, content: encodeSystemTurn(m.content, cloak) }];
     }),
     max_tokens: req.maxTokens ?? 4096,
-    stream: req.stream,
-    ...systemCache.promoted === undefined ? {} : {
-      cache_control: {
-        type: systemCache.promoted.type,
-        ...systemCache.promoted.ttl === undefined ? {} : { ttl: systemCache.promoted.ttl }
+    stream: req.stream
+  };
+  if (systemCache.retarget !== undefined) {
+    const trailing = body.messages.at(-1);
+    if (isRecord(trailing) && Array.isArray(trailing.content)) {
+      for (const block of trailing.content) {
+        if (isRecord(block))
+          delete block.cache_control;
       }
     }
-  };
+    if (!hasVendorCacheControl(req)) {
+      const target = lastCacheableHistoryBlock(body.messages.slice(0, -1));
+      if (target === undefined) {
+        note("anthropic:system-turn-cache-control-dropped");
+      } else if (target.cache_control === undefined) {
+        Object.assign(target, wireCacheControl(systemCache.retarget));
+        note("anthropic:system-turn-cache-control-retargeted");
+      }
+    }
+  }
   if (system !== undefined && system.length > 0)
     body.system = system;
   if (req.temperature !== undefined)
@@ -31024,18 +31065,37 @@ async function vacuum(deps) {
 }
 async function getDatabaseOverview(deps) {
   const live = deps.store.databasePath;
-  const stats = await deps.store.maintenance.stats();
-  const settings = await deps.store.config.getSettings();
+  const [stats, tables, settings] = await Promise.all([
+    deps.store.maintenance.stats(),
+    deps.store.maintenance.tables(),
+    deps.store.config.getSettings()
+  ]);
+  const shared = {
+    engine: deps.store.engine,
+    location: live,
+    stats,
+    logicalBytes: stats.pageSize * stats.pageCount,
+    freePageBytes: stats.pageSize * stats.freelistCount,
+    retention: retentionOf(settings),
+    tables
+  };
+  if (deps.store.engine === "postgres") {
+    return {
+      ...shared,
+      fileBytes: 0,
+      walBytes: 0,
+      bodiesBytes: tables.find((table) => table.name === "request_bodies")?.bytes ?? 0,
+      freeDiskBytes: null,
+      snapshots: { count: 0, totalBytes: 0, latestAt: null }
+    };
+  }
   const snapshots = listSnapshots(deps);
   return {
-    stats,
+    ...shared,
     fileBytes: deps.fs.stat(live)?.size ?? 0,
     walBytes: deps.fs.stat(`${live}-wal`)?.size ?? 0,
     bodiesBytes: deps.fs.dirBytes(bodiesDirFor(live)),
-    logicalBytes: stats.pageSize * stats.pageCount,
-    freePageBytes: stats.pageSize * stats.freelistCount,
     freeDiskBytes: deps.fs.freeBytes(dirname2(live)),
-    retention: retentionOf(settings),
     snapshots: {
       count: snapshots.length,
       totalBytes: snapshots.reduce((sum2, snapshot) => sum2 + snapshot.sizeBytes, 0),
@@ -48520,15 +48580,27 @@ function adminRoutes(deps) {
     const live = await deps.store.maintenance.nodes(deps.now());
     const reads = await Promise.allSettled(live.map((entry) => fleet.read(entry.id, consoleQuery).then((read) => ({ nodeId: entry.id, read }))));
     const lines = [];
-    for (const outcome of reads) {
-      if (outcome.status !== "fulfilled")
+    let captured = false;
+    const unreachable2 = [];
+    for (const [index, outcome] of reads.entries()) {
+      if (outcome.status !== "fulfilled") {
+        unreachable2.push(live[index]?.id ?? "");
         continue;
+      }
+      if (outcome.value.read.source !== "none")
+        captured = true;
       for (const line of outcome.value.read.lines) {
         lines.push({ ...line, nodeId: outcome.value.nodeId });
       }
     }
+    if (unreachable2.length === reads.length) {
+      throw new GatewayError("TIMEOUT", "no process in the fleet answered a console read");
+    }
+    const reach = unreachable2.length === 0 ? {} : { unreachable: unreachable2 };
+    if (!captured)
+      return { source: "none", lines: [], ...reach };
     lines.sort((a, b) => (a.at ?? Number.POSITIVE_INFINITY) - (b.at ?? Number.POSITIVE_INFINITY));
-    return { source: "fleet", lines: lines.slice(-consoleQuery.lines) };
+    return { source: "fleet", lines: lines.slice(-consoleQuery.lines), ...reach };
   }).get("/api/nodes", async ({ request: request2 }) => {
     await requireReader(request2, deps.admin);
     const nodes = await deps.store.maintenance.nodes(deps.now());

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "omnigateway",
-  "version": "0.8.0",
+  "version": "0.8.1",
   "description": "Self-hosted AI gateway with Anthropic- and OpenAI-compatible APIs, an admin console, and a CLI",
   "license": "MIT",
   "author": "Harismawan <mail@harismawan.com>",

package/public/assets/{Chip-DG-mqRx0.js → Chip-BZWa7onX.js}

@@ -1,4 +1,4 @@
-import{m as e}from"./Rack-xBT_6EtM.js";import{jsx as t}from"react/jsx-runtime";import n from"styled-components";var r={neutral:{fg:`var(--ink-dim)`,bg:`transparent`},ok:{fg:`var(--ok)`,bg:`var(--ok-wash)`},warn:{fg:`var(--warn)`,bg:`var(--warn-wash)`},down:{fg:`var(--down)`,bg:`var(--down-wash)`},accent:{fg:`var(--accent)`,bg:`var(--accent-wash)`}},i=n.span`
+import{m as e}from"./Rack-CRuJZY00.js";import{jsx as t}from"react/jsx-runtime";import n from"styled-components";var r={neutral:{fg:`var(--ink-dim)`,bg:`transparent`},ok:{fg:`var(--ok)`,bg:`var(--ok-wash)`},warn:{fg:`var(--warn)`,bg:`var(--warn-wash)`},down:{fg:`var(--down)`,bg:`var(--down-wash)`},accent:{fg:`var(--accent)`,bg:`var(--accent-wash)`}},i=n.span`
   display: inline-flex;
   align-items: center;
   gap: 4px;

package/public/assets/{CopyValue-CItB094V.js → CopyValue-Cs2JUZQD.js}

@@ -1,4 +1,4 @@
-import{C as e,O as t}from"./Panel-CYQvfI3L.js";import{b as n}from"./Rack-xBT_6EtM.js";import{useCallback as r,useEffect as i,useRef as a,useState as o}from"react";import{jsx as s,jsxs as c}from"react/jsx-runtime";import l from"styled-components";var u=n(`check`,[[`path`,{d:`M20 6 9 17l-5-5`,key:`1gmf2c`}]]),d=n(`copy`,[[`rect`,{width:`14`,height:`14`,x:`8`,y:`8`,rx:`2`,ry:`2`,key:`17jyea`}],[`path`,{d:`M4 16c-1.1 0-2-.9-2-2V4c0-1.1.9-2 2-2h10c1.1 0 2 .9 2 2`,key:`zix9uf`}]]);function f(e){if(e===null||typeof getSelection!=`function`)return!1;let t=getSelection();if(t===null)return!1;try{let n=document.createRange();return n.selectNodeContents(e),t.removeAllRanges(),t.addRange(n),!0}catch{return!1}}async function p(e,t){if(typeof navigator<`u`&&navigator.clipboard!==void 0)try{return await navigator.clipboard.writeText(e),`copied`}catch{}let n=f(t??null);if(n&&typeof document.execCommand==`function`)try{if(document.execCommand(`copy`))return`copied`}catch{}return n?`selected`:`failed`}var m=l.div`
+import{C as e,O as t}from"./Panel-CYQvfI3L.js";import{b as n}from"./Rack-CRuJZY00.js";import{useCallback as r,useEffect as i,useRef as a,useState as o}from"react";import{jsx as s,jsxs as c}from"react/jsx-runtime";import l from"styled-components";var u=n(`check`,[[`path`,{d:`M20 6 9 17l-5-5`,key:`1gmf2c`}]]),d=n(`copy`,[[`rect`,{width:`14`,height:`14`,x:`8`,y:`8`,rx:`2`,ry:`2`,key:`17jyea`}],[`path`,{d:`M4 16c-1.1 0-2-.9-2-2V4c0-1.1.9-2 2-2h10c1.1 0 2 .9 2 2`,key:`zix9uf`}]]);function f(e){if(e===null||typeof getSelection!=`function`)return!1;let t=getSelection();if(t===null)return!1;try{let n=document.createRange();return n.selectNodeContents(e),t.removeAllRanges(),t.addRange(n),!0}catch{return!1}}async function p(e,t){if(typeof navigator<`u`&&navigator.clipboard!==void 0)try{return await navigator.clipboard.writeText(e),`copied`}catch{}let n=f(t??null);if(n&&typeof document.execCommand==`function`)try{if(document.execCommand(`copy`))return`copied`}catch{}return n?`selected`:`failed`}var m=l.div`
   display: flex;
   align-items: stretch;
   gap: 0;