t5304: test cleaning pack garbage
[git/git.git] / builtin / gc.c
CommitLineData
6757ada4
JB
1/*
2 * git gc builtin command
3 *
4 * Cleanup unreachable files and optimize the repository.
5 *
6 * Copyright (c) 2007 James Bowes
7 *
8 * Based on git-gc.sh, which is
9 *
10 * Copyright (c) 2006 Shawn O. Pearce
11 */
12
baffc0e7 13#include "builtin.h"
697cc8ef 14#include "lockfile.h"
44c637c8 15#include "parse-options.h"
6757ada4 16#include "run-command.h"
4c5baf02 17#include "sigchain.h"
234587fc 18#include "argv-array.h"
eab3296c 19#include "commit.h"
6757ada4
JB
20
21#define FAILED_RUN "failed to run %s"
22
44c637c8 23static const char * const builtin_gc_usage[] = {
9c9b4f2f 24 N_("git gc [<options>]"),
44c637c8
JB
25 NULL
26};
6757ada4 27
56752391 28static int pack_refs = 1;
62aad184 29static int prune_reflogs = 1;
125f8146 30static int aggressive_depth = 250;
1c192f34 31static int aggressive_window = 250;
2c3c4399 32static int gc_auto_threshold = 6700;
97063974 33static int gc_auto_pack_limit = 50;
9f673f94 34static int detach_auto = 1;
d3154b44 35static const char *prune_expire = "2.weeks.ago";
e3df33bb 36static const char *prune_worktrees_expire = "3.months.ago";
6757ada4 37
234587fc
JK
38static struct argv_array pack_refs_cmd = ARGV_ARRAY_INIT;
39static struct argv_array reflog = ARGV_ARRAY_INIT;
40static struct argv_array repack = ARGV_ARRAY_INIT;
41static struct argv_array prune = ARGV_ARRAY_INIT;
e3df33bb 42static struct argv_array prune_worktrees = ARGV_ARRAY_INIT;
234587fc 43static struct argv_array rerere = ARGV_ARRAY_INIT;
6757ada4 44
4c5baf02
JN
45static char *pidfile;
46
47static void remove_pidfile(void)
48{
49 if (pidfile)
50 unlink(pidfile);
51}
52
53static void remove_pidfile_on_signal(int signo)
54{
55 remove_pidfile();
56 sigchain_pop(signo);
57 raise(signo);
58}
59
09dbb90b
NTND
60static void git_config_date_string(const char *key, const char **output)
61{
62 if (git_config_get_string_const(key, output))
63 return;
64 if (strcmp(*output, "now")) {
65 unsigned long now = approxidate("now");
66 if (approxidate(*output) >= now)
67 git_die_config(key, _("Invalid %s: '%s'"), key, *output);
68 }
69}
70
5801d3b4 71static void gc_config(void)
6757ada4 72{
5801d3b4
TA
73 const char *value;
74
75 if (!git_config_get_value("gc.packrefs", &value)) {
c5e5a2c0 76 if (value && !strcmp(value, "notbare"))
6757ada4
JB
77 pack_refs = -1;
78 else
5801d3b4 79 pack_refs = git_config_bool("gc.packrefs", value);
17815501 80 }
5801d3b4
TA
81
82 git_config_get_int("gc.aggressivewindow", &aggressive_window);
83 git_config_get_int("gc.aggressivedepth", &aggressive_depth);
84 git_config_get_int("gc.auto", &gc_auto_threshold);
85 git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
86 git_config_get_bool("gc.autodetach", &detach_auto);
09dbb90b 87 git_config_date_string("gc.pruneexpire", &prune_expire);
e3df33bb 88 git_config_date_string("gc.pruneworktreesexpire", &prune_worktrees_expire);
5801d3b4 89 git_config(git_default_config, NULL);
6757ada4
JB
90}
91
a087cc98 92static int too_many_loose_objects(void)
2c3c4399
JH
93{
94 /*
95 * Quickly check if a "gc" is needed, by estimating how
96 * many loose objects there are. Because SHA-1 is evenly
97 * distributed, we can check only one and get a reasonable
98 * estimate.
99 */
100 char path[PATH_MAX];
101 const char *objdir = get_object_directory();
102 DIR *dir;
103 struct dirent *ent;
104 int auto_threshold;
105 int num_loose = 0;
106 int needed = 0;
107
17815501
JH
108 if (gc_auto_threshold <= 0)
109 return 0;
110
2c3c4399 111 if (sizeof(path) <= snprintf(path, sizeof(path), "%s/17", objdir)) {
fea6128b 112 warning(_("insanely long object directory %.*s"), 50, objdir);
2c3c4399
JH
113 return 0;
114 }
115 dir = opendir(path);
116 if (!dir)
117 return 0;
118
119 auto_threshold = (gc_auto_threshold + 255) / 256;
120 while ((ent = readdir(dir)) != NULL) {
121 if (strspn(ent->d_name, "0123456789abcdef") != 38 ||
122 ent->d_name[38] != '\0')
123 continue;
124 if (++num_loose > auto_threshold) {
125 needed = 1;
126 break;
127 }
128 }
129 closedir(dir);
130 return needed;
131}
132
17815501
JH
133static int too_many_packs(void)
134{
135 struct packed_git *p;
136 int cnt;
137
138 if (gc_auto_pack_limit <= 0)
139 return 0;
140
141 prepare_packed_git();
142 for (cnt = 0, p = packed_git; p; p = p->next) {
17815501
JH
143 if (!p->pack_local)
144 continue;
01af249f 145 if (p->pack_keep)
17815501
JH
146 continue;
147 /*
148 * Perhaps check the size of the pack and count only
149 * very small ones here?
150 */
151 cnt++;
152 }
153 return gc_auto_pack_limit <= cnt;
154}
155
7e52f566
JK
156static void add_repack_all_option(void)
157{
158 if (prune_expire && !strcmp(prune_expire, "now"))
234587fc 159 argv_array_push(&repack, "-a");
7e52f566 160 else {
234587fc
JK
161 argv_array_push(&repack, "-A");
162 if (prune_expire)
163 argv_array_pushf(&repack, "--unpack-unreachable=%s", prune_expire);
7e52f566
JK
164 }
165}
166
a087cc98
JH
167static int need_to_gc(void)
168{
169 /*
b14d255b
BC
170 * Setting gc.auto to 0 or negative can disable the
171 * automatic gc.
a087cc98 172 */
b14d255b 173 if (gc_auto_threshold <= 0)
95143f9e
JH
174 return 0;
175
17815501
JH
176 /*
177 * If there are too many loose objects, but not too many
178 * packs, we run "repack -d -l". If there are too many packs,
179 * we run "repack -A -d -l". Otherwise we tell the caller
180 * there is no need.
181 */
17815501 182 if (too_many_packs())
7e52f566 183 add_repack_all_option();
17815501
JH
184 else if (!too_many_loose_objects())
185 return 0;
bde30540 186
15048f8a 187 if (run_hook_le(NULL, "pre-auto-gc", NULL))
bde30540 188 return 0;
95143f9e 189 return 1;
a087cc98
JH
190}
191
64a99eb4
NTND
192/* return NULL on success, else hostname running the gc */
193static const char *lock_repo_for_gc(int force, pid_t* ret_pid)
194{
195 static struct lock_file lock;
64a99eb4
NTND
196 char my_host[128];
197 struct strbuf sb = STRBUF_INIT;
198 struct stat st;
199 uintmax_t pid;
200 FILE *fp;
4f1c0b21 201 int fd;
64a99eb4 202
4c5baf02
JN
203 if (pidfile)
204 /* already locked */
205 return NULL;
206
64a99eb4
NTND
207 if (gethostname(my_host, sizeof(my_host)))
208 strcpy(my_host, "unknown");
209
210 fd = hold_lock_file_for_update(&lock, git_path("gc.pid"),
211 LOCK_DIE_ON_ERROR);
212 if (!force) {
4f1c0b21
EP
213 static char locking_host[128];
214 int should_exit;
64a99eb4
NTND
215 fp = fopen(git_path("gc.pid"), "r");
216 memset(locking_host, 0, sizeof(locking_host));
217 should_exit =
218 fp != NULL &&
219 !fstat(fileno(fp), &st) &&
220 /*
221 * 12 hour limit is very generous as gc should
222 * never take that long. On the other hand we
223 * don't really need a strict limit here,
224 * running gc --auto one day late is not a big
225 * problem. --force can be used in manual gc
226 * after the user verifies that no gc is
227 * running.
228 */
229 time(NULL) - st.st_mtime <= 12 * 3600 &&
230 fscanf(fp, "%"PRIuMAX" %127c", &pid, locking_host) == 2 &&
231 /* be gentle to concurrent "gc" on remote hosts */
ed7eda8b 232 (strcmp(locking_host, my_host) || !kill(pid, 0) || errno == EPERM);
64a99eb4
NTND
233 if (fp != NULL)
234 fclose(fp);
235 if (should_exit) {
236 if (fd >= 0)
237 rollback_lock_file(&lock);
238 *ret_pid = pid;
239 return locking_host;
240 }
241 }
242
243 strbuf_addf(&sb, "%"PRIuMAX" %s",
244 (uintmax_t) getpid(), my_host);
245 write_in_full(fd, sb.buf, sb.len);
246 strbuf_release(&sb);
247 commit_lock_file(&lock);
248
4c5baf02
JN
249 pidfile = git_pathdup("gc.pid");
250 sigchain_push_common(remove_pidfile_on_signal);
251 atexit(remove_pidfile);
252
64a99eb4
NTND
253 return NULL;
254}
255
62aad184
NTND
256static int gc_before_repack(void)
257{
258 if (pack_refs && run_command_v_opt(pack_refs_cmd.argv, RUN_GIT_CMD))
259 return error(FAILED_RUN, pack_refs_cmd.argv[0]);
260
261 if (prune_reflogs && run_command_v_opt(reflog.argv, RUN_GIT_CMD))
262 return error(FAILED_RUN, reflog.argv[0]);
263
264 pack_refs = 0;
265 prune_reflogs = 0;
266 return 0;
267}
268
6757ada4
JB
269int cmd_gc(int argc, const char **argv, const char *prefix)
270{
44c637c8 271 int aggressive = 0;
2c3c4399 272 int auto_gc = 0;
a0c14cbb 273 int quiet = 0;
64a99eb4
NTND
274 int force = 0;
275 const char *name;
276 pid_t pid;
6757ada4 277
44c637c8 278 struct option builtin_gc_options[] = {
6705c162
NTND
279 OPT__QUIET(&quiet, N_("suppress progress reporting")),
280 { OPTION_STRING, 0, "prune", &prune_expire, N_("date"),
281 N_("prune unreferenced objects"),
58e9d9d4 282 PARSE_OPT_OPTARG, NULL, (intptr_t)prune_expire },
d5d09d47
SB
283 OPT_BOOL(0, "aggressive", &aggressive, N_("be more thorough (increased runtime)")),
284 OPT_BOOL(0, "auto", &auto_gc, N_("enable auto-gc mode")),
64a99eb4 285 OPT_BOOL(0, "force", &force, N_("force running gc even if there may be another gc running")),
44c637c8
JB
286 OPT_END()
287 };
288
0c8151b6
NTND
289 if (argc == 2 && !strcmp(argv[1], "-h"))
290 usage_with_options(builtin_gc_usage, builtin_gc_options);
291
234587fc
JK
292 argv_array_pushl(&pack_refs_cmd, "pack-refs", "--all", "--prune", NULL);
293 argv_array_pushl(&reflog, "reflog", "expire", "--all", NULL);
294 argv_array_pushl(&repack, "repack", "-d", "-l", NULL);
2cfe2a78 295 argv_array_pushl(&prune, "prune", "--expire", NULL);
df0b6cfb 296 argv_array_pushl(&prune_worktrees, "worktree", "prune", "--expire", NULL);
234587fc
JK
297 argv_array_pushl(&rerere, "rerere", "gc", NULL);
298
5801d3b4 299 gc_config();
6757ada4
JB
300
301 if (pack_refs < 0)
302 pack_refs = !is_bare_repository();
303
37782920
SB
304 argc = parse_options(argc, argv, prefix, builtin_gc_options,
305 builtin_gc_usage, 0);
44c637c8
JB
306 if (argc > 0)
307 usage_with_options(builtin_gc_usage, builtin_gc_options);
308
309 if (aggressive) {
234587fc 310 argv_array_push(&repack, "-f");
125f8146
NTND
311 if (aggressive_depth > 0)
312 argv_array_pushf(&repack, "--depth=%d", aggressive_depth);
234587fc
JK
313 if (aggressive_window > 0)
314 argv_array_pushf(&repack, "--window=%d", aggressive_window);
6757ada4 315 }
a0c14cbb 316 if (quiet)
234587fc 317 argv_array_push(&repack, "-q");
6757ada4 318
2c3c4399
JH
319 if (auto_gc) {
320 /*
321 * Auto-gc should be least intrusive as possible.
322 */
2c3c4399
JH
323 if (!need_to_gc())
324 return 0;
9f673f94
NTND
325 if (!quiet) {
326 if (detach_auto)
327 fprintf(stderr, _("Auto packing the repository in background for optimum performance.\n"));
328 else
329 fprintf(stderr, _("Auto packing the repository for optimum performance.\n"));
330 fprintf(stderr, _("See \"git help gc\" for manual housekeeping.\n"));
331 }
62aad184
NTND
332 if (detach_auto) {
333 if (gc_before_repack())
334 return -1;
9f673f94
NTND
335 /*
336 * failure to daemonize is ok, we'll continue
337 * in foreground
338 */
339 daemonize();
62aad184 340 }
a37cce3b 341 } else
7e52f566 342 add_repack_all_option();
2c3c4399 343
64a99eb4
NTND
344 name = lock_repo_for_gc(force, &pid);
345 if (name) {
346 if (auto_gc)
347 return 0; /* be quiet on --auto */
348 die(_("gc is already running on machine '%s' pid %"PRIuMAX" (use --force if not)"),
349 name, (uintmax_t)pid);
350 }
351
62aad184
NTND
352 if (gc_before_repack())
353 return -1;
6757ada4 354
234587fc
JK
355 if (run_command_v_opt(repack.argv, RUN_GIT_CMD))
356 return error(FAILED_RUN, repack.argv[0]);
6757ada4 357
58e9d9d4 358 if (prune_expire) {
234587fc 359 argv_array_push(&prune, prune_expire);
bf0a59b3 360 if (quiet)
234587fc
JK
361 argv_array_push(&prune, "--no-progress");
362 if (run_command_v_opt(prune.argv, RUN_GIT_CMD))
363 return error(FAILED_RUN, prune.argv[0]);
58e9d9d4 364 }
6757ada4 365
e3df33bb
NTND
366 if (prune_worktrees_expire) {
367 argv_array_push(&prune_worktrees, prune_worktrees_expire);
368 if (run_command_v_opt(prune_worktrees.argv, RUN_GIT_CMD))
369 return error(FAILED_RUN, prune_worktrees.argv[0]);
370 }
371
234587fc
JK
372 if (run_command_v_opt(rerere.argv, RUN_GIT_CMD))
373 return error(FAILED_RUN, rerere.argv[0]);
6757ada4 374
a087cc98 375 if (auto_gc && too_many_loose_objects())
fea6128b
ÆAB
376 warning(_("There are too many unreachable loose objects; "
377 "run 'git prune' to remove them."));
a087cc98 378
6757ada4
JB
379 return 0;
380}