1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543
|
/*
Task Spooler - a task queue system for the unix user
Copyright (C) 2007-2013 LluĂs Batlle i Rossell
Please find the license in the provided COPYING file.
*/
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/select.h>
#ifdef linux
#include <sys/time.h>
#endif
#include <sys/resource.h>
#include <sys/un.h>
#include <errno.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
#include <limits.h>
#include <signal.h>
#include <fcntl.h>
#include <libgen.h>
#include <stdio.h>
#include "main.h"
enum
{
MAXCONN=1000
};
enum Break
{
BREAK,
NOBREAK,
CLOSE
};
/* Prototypes */
static void server_loop(int ls);
static enum Break
client_read(int index);
static void end_server(int ls);
static void s_newjob_ok(int index);
static void s_newjob_nok(int index);
static void s_runjob(int jobid, int index);
static void clean_after_client_disappeared(int socket, int index);
struct Client_conn
{
int socket;
int hasjob;
int jobid;
};
/* Globals */
static struct Client_conn client_cs[MAXCONN];
static int nconnections;
static char *path;
static int max_descriptors;
/* in jobs.c */
extern int max_jobs;
static void s_send_version(int s)
{
struct msg m;
m.type = VERSION;
m.u.version = PROTOCOL_VERSION;
send_msg(s, &m);
}
static void sigterm_handler(int n)
{
const char *dumpfilename;
int fd;
/* Dump the job list if we should to */
dumpfilename = getenv("TS_SAVELIST");
if (dumpfilename != NULL)
{
fd = open(dumpfilename, O_WRONLY | O_APPEND | O_CREAT, 0600);
if (fd != -1)
{
joblist_dump(fd);
close(fd);
} else
warning("The TS_SAVELIST file \"%s\" cannot be opened",
dumpfilename);
}
/* path will be initialized for sure, before installing the handler */
unlink(path);
exit(1);
}
static void set_default_maxslots()
{
char *str;
str = getenv("TS_SLOTS");
if (str != NULL)
{
int slots;
slots = abs(atoi(str));
s_set_max_slots(slots);
}
}
static void install_sigterm_handler()
{
struct sigaction act;
act.sa_handler = sigterm_handler;
/* Reset the mask */
memset(&act.sa_mask,0,sizeof(act.sa_mask));
act.sa_flags = 0;
sigaction(SIGTERM, &act, NULL);
}
static int get_max_descriptors()
{
const int MARGIN = 5; /* stdin, stderr, listen socket, and whatever */
int max;
struct rlimit rlim;
int res;
const char *str;
max = MAXCONN;
str = getenv("TS_MAXCONN");
if (str != NULL)
{
int user_maxconn;
user_maxconn = abs(atoi(str));
if (max > user_maxconn)
max = user_maxconn;
}
if (max > FD_SETSIZE)
max = FD_SETSIZE - MARGIN;
/* I'd like to use OPEN_MAX or NR_OPEN, but I don't know if any
* of them is POSIX compliant */
res = getrlimit(RLIMIT_NOFILE, &rlim);
if (res != 0)
warning("getrlimit for open files");
else
{
if (max > rlim.rlim_cur)
max = rlim.rlim_cur - MARGIN;
}
if (max < 1)
error("Too few opened descriptors available");
return max;
}
void server_main(int notify_fd, char *_path, int ls)
{
char *dirpath;
process_type = SERVER;
max_descriptors = get_max_descriptors();
/* Arbitrary limit, that will block the enqueuing, but should allow space
* for usual ts queries */
max_jobs = max_descriptors - 5;
path = _path;
/* Move the server to the socket directory */
dirpath = malloc(strlen(path)+1);
strcpy(dirpath, path);
chdir(dirname(dirpath));
free(dirpath);
nconnections = 0;
install_sigterm_handler();
set_default_maxslots();
notify_parent(notify_fd);
server_loop(ls);
}
static int get_conn_of_jobid(int jobid)
{
int i;
for(i=0; i< nconnections; ++i)
if (client_cs[i].hasjob && client_cs[i].jobid == jobid)
return i;
return -1;
}
static void server_loop(int ls)
{
fd_set readset;
int i;
int maxfd;
int keep_loop = 1;
int newjob;
while (keep_loop)
{
FD_ZERO(&readset);
maxfd = 0;
/* If we can accept more connections, go on.
* Otherwise, the system block them (no accept will be done). */
if (nconnections < max_descriptors)
{
FD_SET(ls,&readset);
maxfd = ls;
}
for(i=0; i< nconnections; ++i)
{
FD_SET(client_cs[i].socket, &readset);
if (client_cs[i].socket > maxfd)
maxfd = client_cs[i].socket;
}
select(maxfd + 1, &readset, NULL, NULL, NULL);
if (FD_ISSET(ls,&readset))
{
int cs;
cs = accept(ls, NULL, NULL);
if (cs == -1)
error("Accepting from %i", ls);
client_cs[nconnections].hasjob = 0;
client_cs[nconnections].socket = cs;
++nconnections;
}
for(i=0; i< nconnections; ++i)
if (FD_ISSET(client_cs[i].socket, &readset))
{
enum Break b;
b = client_read(i);
/* Check if we should break */
if (b == CLOSE)
{
warning("Closing");
/* On unknown message, we close the client,
or it may hang waiting for an answer */
clean_after_client_disappeared(client_cs[i].socket, i);
}
else if (b == BREAK)
keep_loop = 0;
}
/* This will return firstjob->jobid or -1 */
newjob = next_run_job();
if (newjob != -1)
{
int conn, awaken_job;
conn = get_conn_of_jobid(newjob);
/* This next marks the firstjob state to RUNNING */
s_mark_job_running(newjob);
s_runjob(newjob, conn);
while ((awaken_job = wake_hold_client()) != -1)
{
int wake_conn = get_conn_of_jobid(awaken_job);
if (wake_conn == -1)
error("The job awaken does not have a connection open");
s_newjob_ok(wake_conn);
}
}
}
end_server(ls);
}
static void end_server(int ls)
{
close(ls);
unlink(path);
/* This comes from the parent, in the fork after server_main.
* This is the last use of path in this process.*/
free(path);
}
static void remove_connection(int index)
{
int i;
if(client_cs[index].hasjob)
{
s_removejob(client_cs[index].jobid);
}
for(i=index; i<(nconnections-1); ++i)
{
memcpy(&client_cs[i], &client_cs[i+1], sizeof(client_cs[0]));
}
nconnections--;
}
static void
clean_after_client_disappeared(int socket, int index)
{
/* Act as if the job ended. */
int jobid = client_cs[index].jobid;
if (client_cs[index].hasjob)
{
struct Result r;
r.errorlevel = -1;
r.died_by_signal = 1;
r.signal = SIGKILL;
r.user_ms = 0;
r.system_ms = 0;
r.real_ms = 0;
r.skipped = 0;
warning("JobID %i quit while running.", jobid);
job_finished(&r, jobid);
/* For the dependencies */
check_notify_list(jobid);
/* We don't want this connection to do anything
* more related to the jobid, secially on remove_connection
* when we receive the EOC. */
client_cs[index].hasjob = 0;
}
else
/* If it doesn't have a running job,
* it may well be a notification */
s_remove_notification(socket);
close(socket);
remove_connection(index);
}
static enum Break
client_read(int index)
{
struct msg m;
int s;
int res;
s = client_cs[index].socket;
/* Read the message */
res = recv_msg(s, &m);
if (res == -1)
{
warning("client recv failed");
clean_after_client_disappeared(s, index);
return NOBREAK;
}
else if (res == 0)
{
clean_after_client_disappeared(s, index);
return NOBREAK;
}
/* Process message */
switch(m.type)
{
case KILL_SERVER:
return BREAK; /* break in the parent*/
break;
case NEWJOB:
client_cs[index].jobid = s_newjob(s, &m);
client_cs[index].hasjob = 1;
if (!job_is_holding_client(client_cs[index].jobid))
s_newjob_ok(index);
else if (!m.u.newjob.wait_enqueuing)
{
s_newjob_nok(index);
clean_after_client_disappeared(s, index);
}
break;
case RUNJOB_OK:
{
char *buffer = 0;
if (m.u.output.store_output)
{
/* Receive the output filename */
buffer = (char *) malloc(m.u.output.ofilename_size);
res = recv_bytes(s, buffer,
m.u.output.ofilename_size);
if (res != m.u.output.ofilename_size)
error("Reading the ofilename");
}
s_process_runjob_ok(client_cs[index].jobid, buffer,
m.u.output.pid);
}
break;
case LIST:
s_list(s);
/* We must actively close, meaning End of Lines */
close(s);
remove_connection(index);
break;
case INFO:
s_job_info(s, m.u.jobid);
close(s);
remove_connection(index);
break;
case ENDJOB:
job_finished(&m.u.result, client_cs[index].jobid);
/* For the dependencies */
check_notify_list(client_cs[index].jobid);
/* We don't want this connection to do anything
* more related to the jobid, secially on remove_connection
* when we receive the EOC. */
client_cs[index].hasjob = 0;
break;
case CLEAR_FINISHED:
s_clear_finished();
break;
case ASK_OUTPUT:
s_send_output(s, m.u.jobid);
break;
case REMOVEJOB:
{
int went_ok;
/* Will update the jobid. If it's -1, will set the jobid found */
went_ok = s_remove_job(s, &m.u.jobid);
if (went_ok)
{
int i;
for(i = 0; i < nconnections; ++i)
{
if (client_cs[i].hasjob && client_cs[i].jobid == m.u.jobid)
{
close(client_cs[i].socket);
/* So remove_connection doesn't call s_removejob again */
client_cs[i].hasjob = 0;
/* We don't try to remove any notification related to
* 'i', because it will be for sure a ts client for a job */
remove_connection(i);
}
}
}
}
break;
case WAITJOB:
s_wait_job(s, m.u.jobid);
break;
case WAIT_RUNNING_JOB:
s_wait_running_job(s, m.u.jobid);
break;
case URGENT:
s_move_urgent(s, m.u.jobid);
break;
case SET_MAX_SLOTS:
s_set_max_slots(m.u.max_slots);
break;
case GET_MAX_SLOTS:
s_get_max_slots(s);
break;
case SWAP_JOBS:
s_swap_jobs(s, m.u.swap.jobid1,
m.u.swap.jobid2);
break;
case GET_STATE:
s_send_state(s, m.u.jobid);
break;
case GET_VERSION:
s_send_version(s);
break;
default:
/* Command not supported */
/* On unknown message, we close the client,
or it may hang waiting for an answer */
warning("Unknown message: %i", m.type);
return CLOSE;
}
return NOBREAK; /* normal */
}
static void s_runjob(int jobid, int index)
{
int s;
if (!client_cs[index].hasjob)
error("Run job of the client %i which doesn't have any job", index);
s = client_cs[index].socket;
s_send_runjob(s, jobid);
}
static void s_newjob_ok(int index)
{
int s;
struct msg m;
if (!client_cs[index].hasjob)
error("Run job of the client %i which doesn't have any job", index);
s = client_cs[index].socket;
m.type = NEWJOB_OK;
m.u.jobid = client_cs[index].jobid;
send_msg(s, &m);
}
static void s_newjob_nok(int index)
{
int s;
struct msg m;
if (!client_cs[index].hasjob)
error("Run job of the client %i which doesn't have any job", index);
s = client_cs[index].socket;
m.type = NEWJOB_NOK;
send_msg(s, &m);
}
static void dump_conn_struct(FILE *out, const struct Client_conn *p)
{
fprintf(out, " new_conn\n");
fprintf(out, " socket %i\n", p->socket);
fprintf(out, " hasjob \"%i\"\n", p->hasjob);
fprintf(out, " jobid %i\n", p->jobid);
}
void dump_conns_struct(FILE *out)
{
int i;
fprintf(out, "New_conns");
for(i=0; i < nconnections; ++i)
{
dump_conn_struct(out, &client_cs[i]);
}
}
|