milestone 28: query provenance — every logged query is exactly explainable
Gates / frontend (push) Successful in 1m36s
Gates / test (push) Successful in 1m56s
Gates / test-aarch64 (push) Successful in 7m37s
Gates / package (push) Successful in 9m12s
Gates / container (push) Successful in 13s
CI / gates (push) Successful in 19m4s

query rows gain qclass, rcode, group, policy action and reason, the
matched rule or list entry with its source, cname and safe-search
targets, route kind, forward zone, and the resolver that actually
answered — the pool and local markers die. servfails are logged and
name the resolver that lost; post-parse protocol refusals become rows.
a detail page at /queries/:id renders the ordered explanation, and
coverage watermarks distinguish an empty history from a missing one.

the schema fingerprint changes: existing query history is recreated
with the old file kept aside and the reset filed as a resolved
diagnostic. fixes an oversized udp reply being rebuilt as noerror,
which handed clients a truncated nxdomain as success.
This commit is contained in:
2026-08-22 09:16:40 +02:00
parent 7e6cb507d2
commit 0fd6bbd312
65 changed files with 7036 additions and 685 deletions
+148 -10
View File
@@ -1,4 +1,5 @@
//! `GET /api/queries` — the query log, newest first (ruling 11).
//! `GET /api/queries` — the query log, newest first (ruling 11) — and
//! `GET /api/queries/{id}`, one row of it fully explained.
//!
//! Keyset pagination rather than an offset: the table is append-only and the
//! UI reads the head of it, so `id < before` is one index seek no matter how
@@ -13,9 +14,11 @@
const std = @import("std");
const Allocator = std.mem.Allocator;
const coverage = @import("../coverage.zig");
const db = @import("../../storage/db.zig");
const http_util = @import("../http_util.zig");
const logger = @import("../../storage/logger.zig");
const provenance_view = @import("../provenance_view.zig");
const queries_repo = @import("../../storage/repositories/queries_repo.zig");
const server = @import("../server.zig");
@@ -41,6 +44,10 @@ pub const Page = struct {
queries: []const queries_repo.QueryRow,
/// The cursor for the next page, or null when this page is the last one.
next_before: ?i64,
/// Whether the log still covers the window the filter asked for. A client
/// that reads rows without reading this cannot tell an empty window from a
/// pruned one.
coverage: coverage.Coverage,
};
pub const FilterError = error{
@@ -110,6 +117,7 @@ pub fn page(
return .{
.queries = rows.items,
.next_before = if (full and rows.items.len != 0) rows.items[rows.items.len - 1].id else null,
.coverage = try coverage.read(database, filter.since),
};
}
@@ -138,10 +146,37 @@ pub fn list(
return http_util.respondJson(request, .ok, result, &.{});
}
/// `GET /api/queries/{id}` — one query, fully explained.
///
/// The row is the whole answer: every field is a fact recorded when the query
/// was answered, so nothing here is joined against current configuration. A
/// group or blocklist renamed since keeps the name it had.
pub fn detail(
state: *server.WebState,
io: std.Io,
request: *http_util.Request,
) http_util.HandlerError!void {
_ = io;
const database = state.querylog_db orelse
return http_util.respondError(request, .service_unavailable, "query log unavailable");
const row = queries_repo.detailById(database, request.arena, request.id.?) catch |err| {
log.warn("query log read failed: {s}", .{@errorName(err)});
return http_util.respondError(request, .internal_server_error, "internal error");
};
// An id retention has pruned and one that never existed are the same
// answer, and the API does not pretend to tell them apart.
const found = row orelse return http_util.respondError(request, .not_found, "not found");
return http_util.respondJson(request, .ok, provenance_view.fromDetail(found), &.{});
}
// ---------------------------------------------------------------------------
// tests
// ---------------------------------------------------------------------------
const provenance = @import("../../storage/provenance.zig");
const querylog_schema = @import("../../storage/querylog_schema.zig");
const testing = std.testing;
@@ -210,21 +245,40 @@ fn openLog() !db.Db {
return database;
}
/// Rows the resolver could actually have written. Ruling 20 ties the three
/// route facts together: a block never consulted the cache, so its `cache_hit`
/// is NULL rather than false; a cache hit has no upstream to name; and only an
/// upstream answer carries one. A fixture that broke those ties would let a
/// serializer regression pass here and fail on real rows.
fn seed(database: *db.Db, count: usize) !void {
var writer = try queries_repo.BatchWriter.init(database);
defer writer.deinit();
var rows: [16]queries_repo.Row = undefined;
for (rows[0..count], 0..) |*row, i| {
const blocked = i % 2 == 0;
const from_cache = i % 4 == 1;
row.* = .{
.timestamp = 1_700_000_000 + @as(i64, @intCast(i)),
.domain = if (i % 2 == 0) "ads.example" else "safe.example",
.domain = if (blocked) "ads.example" else "safe.example",
.client_ip = "192.0.2.10",
.qtype = 1,
.blocked = i % 2 == 0,
.block_reason = if (i % 2 == 0) "blocklist_domain" else null,
.qclass = 1,
.rcode = 0,
.blocked = blocked,
.response_time_us = 500,
.cache_hit = false,
.upstream = null,
.cache_hit = if (blocked) null else from_cache,
.upstream = if (blocked or from_cache) null else "9.9.9.9",
.group_id = 1,
.group_name = "default",
.policy_action = if (blocked) .block else .allow,
.policy_reason = if (blocked) .blocklist_domain else .no_match,
.matched = if (blocked) "ads.example" else null,
.source_id = null,
.source_name = null,
.cname_target = null,
.safe_search_target = null,
.route_kind = if (blocked) .blocked else if (from_cache) .cache else .upstream,
.forward_zone = null,
};
}
try writer.writeBatch(rows[0..count]);
@@ -297,6 +351,37 @@ test "the parsed filters narrow the rows the page returns" {
try testing.expectEqual(@as(usize, 0), nobody.queries.len);
}
test "the seeded rows carry only the route shapes ruling 20 allows" {
var database = try openLog();
defer database.close();
try seed(&database, 4);
var arena: std.heap.ArenaAllocator = .init(testing.allocator);
defer arena.deinit();
var seen: std.EnumSet(provenance.RouteKind) = .initEmpty();
for ((try page(&database, arena.allocator(), .{})).queries) |row| {
seen.insert(row.route_kind);
switch (row.route_kind) {
.blocked => {
try testing.expectEqual(@as(?bool, null), row.cache_hit);
try testing.expectEqualStrings("", row.upstream);
},
.cache => {
try testing.expectEqual(@as(?bool, true), row.cache_hit);
try testing.expectEqualStrings("", row.upstream);
},
.upstream => {
try testing.expectEqual(@as(?bool, false), row.cache_hit);
try testing.expectEqualStrings("9.9.9.9", row.upstream);
},
.local, .forward_zone, .rejected => return error.UnseededRouteKind,
}
}
// All three, so the serializer tests below read every shape the fixture claims.
try testing.expectEqual(@as(usize, 3), seen.count());
}
test "the page serializes as the envelope ruling 11 defines" {
var database = try openLog();
defer database.close();
@@ -312,14 +397,67 @@ test "the page serializes as the envelope ruling 11 defines" {
const text = allocating.written();
try testing.expect(std.mem.startsWith(u8, text, "{\"queries\":["));
try testing.expect(std.mem.endsWith(u8, text, "\"next_before\":null}"));
for ([_][]const u8{
"\"id\":", "\"ts\":", "\"domain\":", "\"client_ip\":",
"\"qtype\":", "\"blocked\":", "\"cache_hit\":", "\"upstream\":",
"\"upstream\":", "\"response_time_us\":", "\"block_reason\":",
"\"id\":", "\"ts\":", "\"domain\":", "\"client_ip\":",
"\"qtype\":", "\"qclass\":", "\"rcode\":", "\"blocked\":",
"\"cache_hit\":", "\"upstream\":", "\"response_time_us\":", "\"policy_action\":",
"\"policy_reason\":", "\"route_kind\":",
}) |field| {
try testing.expect(std.mem.containsAtLeast(u8, text, 1, field));
}
// W1's ruling: a NULL column reads as "", and "" stays "" on the wire.
try testing.expect(std.mem.containsAtLeast(u8, text, 1, "\"upstream\":\"\""));
// The column the provenance columns replaced.
try testing.expect(!std.mem.containsAtLeast(u8, text, 1, "block_reason"));
try testing.expect(std.mem.containsAtLeast(u8, text, 1, "\"coverage\":{\"complete\":"));
}
test "the coverage of a page answers the window the filter asked for" {
var database = try openLog();
defer database.close();
try seed(&database, 1);
var arena: std.heap.ArenaAllocator = .init(testing.allocator);
defer arena.deinit();
const watermark = try queries_repo.availableSince(&database);
const unbounded = try page(&database, arena.allocator(), .{});
try testing.expectEqual(watermark, unbounded.coverage.available_since);
try testing.expect(!unbounded.coverage.complete);
const covered = try page(&database, arena.allocator(), .{ .since = watermark });
try testing.expect(covered.coverage.complete);
const older = try page(&database, arena.allocator(), .{ .since = watermark - 1 });
try testing.expect(!older.coverage.complete);
}
test "a detail row carries every provenance field the row stored" {
var database = try openLog();
defer database.close();
try seed(&database, 1);
var arena_state: std.heap.ArenaAllocator = .init(testing.allocator);
defer arena_state.deinit();
const arena = arena_state.allocator();
const row = (try queries_repo.detailById(&database, arena, 1)).?;
const view = provenance_view.fromDetail(row);
try testing.expectEqual(@as(i64, 1), view.id);
try testing.expectEqualStrings("ads.example", view.request.domain);
try testing.expectEqualStrings("192.0.2.10", view.request.client);
try testing.expectEqual(@as(u16, 1), view.request.qclass);
try testing.expectEqualStrings("default", view.group.name);
try testing.expectEqual(provenance.PolicyAction.block, view.policy.action);
try testing.expectEqualStrings("ads.example", view.policy.matched);
try testing.expectEqual(provenance.RouteKind.blocked, view.route.kind);
try testing.expectEqualStrings("", view.route.upstream);
try testing.expectEqual(@as(?i64, 500), view.response.duration_us);
try testing.expectEqual(
@as(?queries_repo.QueryDetail, null),
try queries_repo.detailById(&database, arena, 99),
);
}