carlos / platform Public

Download this file

1	package host
2
3	import (
4	"context"
5	"os"
6	"path/filepath"
7	"runtime"
8	"strings"
9	"testing"
10	)
11
12	func hostexec(t *testing.T, f *applyFixture) string {
13	t.Helper()
14	dir := filepath.Join(f.a.dataDir, hostexecDir)
15	if err := os.MkdirAll(dir, 0o755); err != nil {
16	t.Fatal(err)
17	}
18	return dir
19	}
20
21	func assertNotUpdated(t *testing.T, f *applyFixture, why string) {
22	t.Helper()
23	r := f.result()
24	if r == nil || r.Outcome != systemOutcomeDeferred || !strings.Contains(r.Reason, why) {
25	t.Fatalf("result = %+v, want deferred naming %q", r, why)
26	}
27	if r.terminal() {
28	t.Fatal("the agent would never retry this promote")
29	}
30	if bin, _ := os.ReadFile(systemBinaryPath(f.a.optDir)); !strings.Contains(string(bin), "carlos v1") {
31	t.Fatal("the binary changed under a refusal")
32	}
33	if j, _ := readSystemJournal(f.a.optDir); j != nil {
34	t.Fatalf("a refusal opened a journal: %+v", j)
35	}
36	if _, err := os.Stat(updateBegunPath(f.a.optDir)); !os.IsNotExist(err) {
37	t.Fatalf("the update-begun marker outlived the refusal: %v", err)
38	}
39	if f.called("systemctl stop") || f.called("systemctl restart") {
40	t.Fatal("a refused update touched a unit")
41	}
42	}
43
44	// Spec §5.10: any controller record, in any state, refuses the update.
45	func TestApplyRefusesWhileThisBoxTakesPartInAClaimOperation(t *testing.T) {
46	t.Parallel()
47	for _, state := range []string{"pending", "active", "lost"} {
48	t.Run(state, func(t *testing.T) {
49	f, _ := applyHappy(t)
50	f.a.regionHA = true
51	os.WriteFile(filepath.Join(hostexec(t, f), "shop.bab.test.json"), []byte(`{"state":"`+state+`"}`), 0o600)
52	if err := f.a.Run(context.Background(), false); err != nil {
53	t.Fatal(err)
54	}
55	assertNotUpdated(t, f, "takes part in a running claim operation")
56	})
57	}
58	}
59
60	// Review Focus 4: what root cannot read is a refusal, never a pass.
61	func TestApplyRefusesWhenItCannotTellWhetherThisBoxTakesPart(t *testing.T) {
62	t.Parallel()
63	cases := map[string]func(t *testing.T, f *applyFixture){
64	"a temporary file": func(t *testing.T, f *applyFixture) {
65	os.WriteFile(filepath.Join(hostexec(t, f), ".shop.bab.test.json.tmp"), nil, 0o600)
66	},
67	"garbage": func(t *testing.T, f *applyFixture) {
68	os.WriteFile(filepath.Join(hostexec(t, f), "shop.bab.test.json"), []byte("{half"), 0o600)
69	},
70	"a symlink in place of the directory": func(t *testing.T, f *applyFixture) {
71	elsewhere := t.TempDir()
72	if err := os.Symlink(elsewhere, filepath.Join(f.a.dataDir, hostexecDir)); err != nil {
73	t.Skip("no symlinks here:", err)
74	}
75	},
76	"a file in place of the directory": func(t *testing.T, f *applyFixture) {
77	os.WriteFile(filepath.Join(f.a.dataDir, hostexecDir), nil, 0o600)
78	},
79	}
80	for name, setup := range cases {
81	t.Run(name, func(t *testing.T) {
82	f, _ := applyHappy(t)
83	f.a.regionHA = true
84	setup(t, f)
85	if err := f.a.Run(context.Background(), false); err != nil {
86	t.Fatal(err)
87	}
88	assertNotUpdated(t, f, "hostexec")
89	})
90	}
91	}
92
93	func TestApplyRefusesWhenItCannotListTheControllerDirectory(t *testing.T) {
94	if runtime.GOOS == "windows" || os.Geteuid() == 0 {
95	t.Skip("permission bits do not stop this user")
96	}
97	f, _ := applyHappy(t)
98	f.a.regionHA = true
99	dir := hostexec(t, f)
100	os.Chmod(dir, 0)
101	t.Cleanup(func() { os.Chmod(dir, 0o755) })
102	if err := f.a.Run(context.Background(), false); err != nil {
103	t.Fatal(err)
104	}
105	assertNotUpdated(t, f, "cannot be listed")
106	}
107
108	// The marker is written before the box changes and is gone once the
109	// update is done; an empty controller directory is no participation.
110	func TestApplyHoldsTheUpdateBegunMarkerAcrossTheInstall(t *testing.T) {
111	f, req := applyHappy(t)
112	f.a.regionHA = true
113	hostexec(t, f) // present and empty
114	seen := false
115	orig := f.a.ops.systemctl
116	f.a.ops.systemctl = func(args ...string) (string, error) {
117	if args[0] == "restart" && args[len(args)-1] == systemEdgeUnit {
118	_, err := os.Stat(updateBegunPath(f.a.optDir))
119	seen = err == nil
120	}
121	return orig(args...)
122	}
123	if err := f.a.Run(context.Background(), false); err != nil {
124	t.Fatal(err)
125	}
126	if r := f.result(); r == nil || r.Outcome != systemOutcomeApplied || !r.matches("stable", "v2", req.PromotedAt) {
127	t.Fatalf("result = %+v", r)
128	}
129	if !seen {
130	t.Fatal("no update-begun marker while the edge restarted onto the new binary")
131	}
132	if _, err := os.Stat(updateBegunPath(f.a.optDir)); !os.IsNotExist(err) {
133	t.Fatalf("marker left after the update finished: %v", err)
134	}
135	}
136
137	func TestApplyWithoutRegionHAWritesNoMarkerAndIgnoresControllerRecords(t *testing.T) {
138	f, _ := applyHappy(t)
139	os.WriteFile(filepath.Join(hostexec(t, f), "shop.bab.test.json"), []byte(`{}`), 0o600)
140	if err := f.a.Run(context.Background(), false); err != nil {
141	t.Fatal(err)
142	}
143	if r := f.result(); r == nil || r.Outcome != systemOutcomeApplied {
144	t.Fatalf("result = %+v: the flag is off, so nothing changes", r)
145	}
146	if _, err := os.Stat(systemGuardDir(f.a.optDir)); !os.IsNotExist(err) {
147	t.Fatalf("the guard directory was created with the flag off: %v", err)
148	}
149	}
150
151	// A crash between the marker and the journal leaves a marker with no
152	// update behind it; the next run, --recover included, clears it.
153	func TestAStaleUpdateBegunMarkerIsClearedByTheNextRun(t *testing.T) {
154	f := newApplyFixture(t, "stable", "v1")
155	os.MkdirAll(systemGuardDir(f.a.optDir), 0o755)
156	os.WriteFile(updateBegunPath(f.a.optDir), []byte(`{"target":"v2"}`), 0o644)
157	if err := f.a.Run(context.Background(), true); err != nil {
158	t.Fatal(err)
159	}
160	if _, err := os.Stat(updateBegunPath(f.a.optDir)); !os.IsNotExist(err) {
161	t.Fatalf("stale marker kept: %v", err)
162	}
163	}
164
165	// A rollback that could not finish keeps its journal for the next run, and
166	// the marker with it: the update is not over.
167	func TestTheMarkerStaysWhileAJournalIsOpen(t *testing.T) {
168	f := newApplyFixture(t, "stable", "v1")
169	f.a.regionHA = true
170	os.MkdirAll(systemGuardDir(f.a.optDir), 0o755)
171	os.WriteFile(updateBegunPath(f.a.optDir), []byte(`{"target":"v2"}`), 0o644)
172	// A journal recovery cannot act on: no carlos.prev to restore.
173	if err := writeJSONDurable(f.a.file("journal.json"), systemJournal{Phase: systemPhaseInstalled, Target: "v2"}, 0o644); err != nil {
174	t.Fatal(err)
175	}
176	_ = f.a.Run(context.Background(), true)
177	if j, _ := readSystemJournal(f.a.optDir); j == nil {
178	t.Fatal("recovery closed the journal; this test needs one it cannot finish")
179	}
180	if _, err := os.Stat(updateBegunPath(f.a.optDir)); err != nil {
181	t.Fatalf("marker cleared while the journal is still open: %v", err)
182	}
183	}
184
185	// Astra plan R1-3: participation defers, it does not refuse. Once the
186	// operation ends and its record is cleared, the same promote applies. The
187	// agent's side, re-requesting a deferred apply, is
188	// TestSystemLegReusesAVerifiedCacheOnARetry.
189	func TestApplyDeferredForParticipationAppliesOnceItEnds(t *testing.T) {
190	f, req := applyHappy(t)
191	f.a.regionHA = true
192	rec := filepath.Join(hostexec(t, f), "shop.bab.test.json")
193	os.WriteFile(rec, []byte(`{"state":"active"}`), 0o600)
194	if err := f.a.Run(context.Background(), false); err != nil {
195	t.Fatal(err)
196	}
197	if r := f.result(); r == nil || r.Outcome != systemOutcomeDeferred || r.terminal() {
198	t.Fatalf("result = %+v, want deferred, which the agent retries", r)
199	}
200	os.Remove(rec)
201	if err := f.a.Run(context.Background(), false); err != nil {
202	t.Fatal(err)
203	}
204	if r := f.result(); r == nil || r.Outcome != systemOutcomeApplied || !r.matches("stable", "v2", req.PromotedAt) {
205	t.Fatalf("result after the operation ended = %+v, want applied", r)
206	}
207	}
208