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