2 * Job control and execution for Storage Daemon
10 Copyright (C) 2000-2005 Kern Sibbald
12 This program is free software; you can redistribute it and/or
13 modify it under the terms of the GNU General Public License as
14 published by the Free Software Foundation; either version 2 of
15 the License, or (at your option) any later version.
17 This program is distributed in the hope that it will be useful,
18 but WITHOUT ANY WARRANTY; without even the implied warranty of
19 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 General Public License for more details.
22 You should have received a copy of the GNU General Public
23 License along with this program; if not, write to the Free
24 Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
32 /* Imported variables */
33 extern uint32_t VolSessionTime;
35 /* Imported functions */
36 extern uint32_t newVolSessionId();
38 /* Forward referenced functions */
39 static bool use_storage_cmd(JCR *jcr);
41 /* Requests from the Director daemon */
42 static char jobcmd[] = "JobId=%d job=%127s job_name=%127s client_name=%127s "
43 "type=%d level=%d FileSet=%127s NoAttr=%d SpoolAttr=%d FileSetMD5=%127s "
44 "SpoolData=%d WritePartAfterJob=%d NewVol=%d\n";
45 static char use_storage[] = "use storage=%127s media_type=%127s "
46 "pool_name=%127s pool_type=%127s append=%d copy=%d stripe=%d\n";
47 static char use_device[] = "use device=%127s\n";
48 //static char query_device[] = "query device=%127s";
51 /* Responses sent to Director daemon */
52 static char OKjob[] = "3000 OK Job SDid=%u SDtime=%u Authorization=%s\n";
53 static char OK_device[] = "3000 OK use device device=%s\n";
54 static char NO_device[] = "3924 Device \"%s\" not in SD Device resources.\n";
55 static char NOT_open[] = "3925 Device \"%s\" could not be opened or does not exist.\n";
56 static char BAD_use[] = "3913 Bad use command: %s\n";
57 static char BAD_job[] = "3915 Bad Job command: %s\n";
58 //static char OK_query[] = "3001 OK query\n";
59 //static char NO_query[] = "3918 Query failed\n";
60 //static char BAD_query[] = "3917 Bad query command: %s\n";
63 * Director requests us to start a job
64 * Basic tasks done here:
65 * - We pickup the JobId to be run from the Director.
66 * - We pickup the device, media, and pool from the Director
67 * - Wait for a connection from the File Daemon (FD)
68 * - Accept commands from the FD (i.e. run the job)
69 * - Return when the connection is terminated or
72 bool job_cmd(JCR *jcr)
76 BSOCK *dir = jcr->dir_bsock;
77 POOL_MEM job_name, client_name, job, fileset_name, fileset_md5;
78 int JobType, level, spool_attributes, no_attributes, spool_data;
79 int write_part_after_job, NewVol;
84 * Get JobId and permissions from Director
86 Dmsg1(100, "<dird: %s\n", dir->msg);
87 if (sscanf(dir->msg, jobcmd, &JobId, job.c_str(), job_name.c_str(),
89 &JobType, &level, fileset_name.c_str(), &no_attributes,
90 &spool_attributes, fileset_md5.c_str(), &spool_data,
91 &write_part_after_job, &NewVol) != 13) {
92 pm_strcpy(jcr->errmsg, dir->msg);
93 bnet_fsend(dir, BAD_job, jcr->errmsg);
94 Dmsg1(100, ">dird: %s\n", dir->msg);
95 Emsg1(M_FATAL, 0, _("Bad Job Command from Director: %s\n"), jcr->errmsg);
96 set_jcr_job_status(jcr, JS_ErrorTerminated);
100 * Since this job could be rescheduled, we
101 * check to see if we have it already. If so
102 * free the old jcr and use the new one.
104 ojcr = get_jcr_by_full_name(job.c_str());
105 if (ojcr && !ojcr->authenticated) {
106 Dmsg2(100, "Found ojcr=0x%x Job %s\n", (unsigned)(long)ojcr, job.c_str());
110 jcr->VolSessionId = newVolSessionId();
111 jcr->VolSessionTime = VolSessionTime;
112 bstrncpy(jcr->Job, job, sizeof(jcr->Job));
113 unbash_spaces(job_name);
114 jcr->job_name = get_pool_memory(PM_NAME);
115 pm_strcpy(jcr->job_name, job_name);
116 unbash_spaces(client_name);
117 jcr->client_name = get_pool_memory(PM_NAME);
118 pm_strcpy(jcr->client_name, client_name);
119 unbash_spaces(fileset_name);
120 jcr->fileset_name = get_pool_memory(PM_NAME);
121 pm_strcpy(jcr->fileset_name, fileset_name);
122 jcr->JobType = JobType;
123 jcr->JobLevel = level;
124 jcr->no_attributes = no_attributes;
125 jcr->spool_attributes = spool_attributes;
126 jcr->spool_data = spool_data;
127 jcr->write_part_after_job = write_part_after_job;
128 jcr->fileset_md5 = get_pool_memory(PM_NAME);
129 pm_strcpy(jcr->fileset_md5, fileset_md5);
130 jcr->NewVolEachJob = NewVol;
132 jcr->authenticated = false;
135 * Pass back an authorization key for the File daemon
137 make_session_key(auth_key, NULL, 1);
138 bnet_fsend(dir, OKjob, jcr->VolSessionId, jcr->VolSessionTime, auth_key);
139 Dmsg1(100, ">dird: %s", dir->msg);
140 jcr->sd_auth_key = bstrdup(auth_key);
141 memset(auth_key, 0, sizeof(auth_key));
142 generate_daemon_event(jcr, "JobStart");
146 bool use_cmd(JCR *jcr)
149 * Wait for the device, media, and pool information
151 if (!use_storage_cmd(jcr)) {
152 set_jcr_job_status(jcr, JS_ErrorTerminated);
153 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
159 bool run_cmd(JCR *jcr)
163 struct timespec timeout;
166 Dmsg1(100, "Run_cmd: %s\n", jcr->dir_bsock->msg);
167 /* The following jobs don't need the FD */
168 switch (jcr->JobType) {
172 jcr->authenticated = true;
177 set_jcr_job_status(jcr, JS_WaitFD); /* wait for FD to connect */
178 dir_send_job_status(jcr);
180 gettimeofday(&tv, &tz);
181 timeout.tv_nsec = tv.tv_usec * 1000;
182 timeout.tv_sec = tv.tv_sec + 30 * 60; /* wait 30 minutes */
184 Dmsg1(100, "%s waiting on FD to contact SD\n", jcr->Job);
186 * Wait for the File daemon to contact us to start the Job,
187 * when he does, we will be released, unless the 30 minutes
191 for ( ;!job_canceled(jcr); ) {
192 errstat = pthread_cond_timedwait(&jcr->job_start_wait, &jcr->mutex, &timeout);
193 if (errstat == 0 || errstat == ETIMEDOUT) {
199 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
201 if (jcr->authenticated && !job_canceled(jcr)) {
202 Dmsg1(100, "Running job %s\n", jcr->Job);
203 run_job(jcr); /* Run the job */
209 * After receiving a connection (in job.c) if it is
210 * from the File daemon, this routine is called.
212 void handle_filed_connection(BSOCK *fd, char *job_name)
216 bmicrosleep(0, 50000); /* wait 50 millisecs */
217 if (!(jcr=get_jcr_by_full_name(job_name))) {
218 Jmsg1(NULL, M_FATAL, 0, _("Job name not found: %s\n"), job_name);
219 Dmsg1(100, "Job name not found: %s\n", job_name);
223 jcr->file_bsock = fd;
224 jcr->file_bsock->jcr = jcr;
226 Dmsg1(110, "Found Job %s\n", job_name);
228 if (jcr->authenticated) {
229 Jmsg2(jcr, M_FATAL, 0, "Hey!!!! JobId %u Job %s already authenticated.\n",
230 jcr->JobId, jcr->Job);
236 * Authenticate the File daemon
238 if (jcr->authenticated || !authenticate_filed(jcr)) {
239 Dmsg1(100, "Authentication failed Job %s\n", jcr->Job);
240 Jmsg(jcr, M_FATAL, 0, _("Unable to authenticate File daemon\n"));
242 jcr->authenticated = true;
243 Dmsg1(110, "OK Authentication Job %s\n", jcr->Job);
247 if (!jcr->authenticated) {
248 set_jcr_job_status(jcr, JS_ErrorTerminated);
250 pthread_cond_signal(&jcr->job_start_wait); /* wake waiting job */
258 * Use Device command from Director
259 * He tells is what Device Name to use, the Media Type,
260 * the Pool Name, and the Pool Type.
262 * Ensure that the device exists and is opened, then store
263 * the media and pool info in the JCR.
268 char name[MAX_NAME_LENGTH];
269 char media_type[MAX_NAME_LENGTH];
270 char pool_name[MAX_NAME_LENGTH];
271 char pool_type[MAX_NAME_LENGTH];
274 static bool use_storage_cmd(JCR *jcr)
276 POOL_MEM store_name, dev_name, media_type, pool_name, pool_type;
277 BSOCK *dir = jcr->dir_bsock;
279 AUTOCHANGER *changer;
288 * If there are multiple devices, the director sends us
289 * use_device for each device that it wants to use.
291 Dmsg1(100, "<dird: %s", dir->msg);
292 dirstore = New(alist(10, not_owned_by_alist));
294 ok = sscanf(dir->msg, use_storage, store_name.c_str(),
295 media_type.c_str(), pool_name.c_str(),
296 pool_type.c_str(), &append, &Copy, &Stripe) == 7;
300 unbash_spaces(store_name);
301 unbash_spaces(media_type);
302 unbash_spaces(pool_name);
303 unbash_spaces(pool_type);
304 store = new DIRSTORE;
305 dirstore->append(store);
306 memset(store, 0, sizeof(DIRSTORE));
307 store->device = New(alist(10));
308 bstrncpy(store->name, store_name, sizeof(store->name));
309 bstrncpy(store->media_type, media_type, sizeof(store->media_type));
310 bstrncpy(store->pool_name, pool_name, sizeof(store->pool_name));
311 bstrncpy(store->pool_type, pool_type, sizeof(store->pool_type));
313 /* Now get all devices */
314 while (bnet_recv(dir) >= 0) {
315 ok = sscanf(dir->msg, use_device, dev_name.c_str()) == 1;
319 unbash_spaces(dev_name);
320 store->device->append(bstrdup(dev_name.c_str()));
322 } while (ok && bnet_recv(dir) >= 0);
324 /* This loop is debug code and can be removed */
325 /* ***FIXME*** turn off in production */
326 foreach_alist(store, dirstore) {
327 Dmsg4(100, "Storage=%s media_type=%s pool=%s pool_type=%s\n",
328 store->name, store->media_type, store->pool_name,
330 foreach_alist(device_name, store->device) {
331 Dmsg1(100, " Device=%s\n", device_name);
338 store = (DIRSTORE *)dirstore->first();
339 foreach_alist(device_name, store->device) {
340 foreach_res(device, R_DEVICE) {
341 /* Find resource, and make sure we were able to open it */
342 if (fnmatch(device_name, device->hdr.name, 0) == 0 &&
343 strcmp(device->media_type, store->media_type) == 0) {
344 const int name_len = MAX_NAME_LENGTH;
348 device->dev = init_dev(jcr, NULL, device);
351 Jmsg(jcr, M_WARNING, 0, _("\n"
352 " Device \"%s\" requested by DIR could not be opened or does not exist.\n"),
354 bnet_fsend(dir, NOT_open, dev_name.c_str());
355 Dmsg1(100, ">dird: %s\n", dir->msg);
358 dcr = new_dcr(jcr, device->dev);
360 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
361 Dmsg1(100, ">dird: %s\n", dir->msg);
364 Dmsg1(100, "Found device %s\n", device->hdr.name);
365 bstrncpy(dcr->pool_name, store->pool_name, name_len);
366 bstrncpy(dcr->pool_type, store->pool_type, name_len);
367 bstrncpy(dcr->media_type, store->media_type, name_len);
368 bstrncpy(dcr->dev_name, device_name, name_len);
370 dcr->Stripe = Stripe;
372 if (append == SD_APPEND) {
373 ok = reserve_device_for_append(dcr);
375 ok = reserve_device_for_read(dcr);
378 bnet_fsend(dir, _("3927 Could not reserve device: %s\n"), dev_name.c_str());
379 Dmsg1(100, ">dird: %s\n", dir->msg);
383 Dmsg1(220, "Got: %s", dir->msg);
384 bash_spaces(dev_name);
385 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
386 Dmsg1(100, ">dird: %s\n", dir->msg);
390 foreach_res(changer, R_AUTOCHANGER) {
391 /* Find resource, and make sure we were able to open it */
392 if (fnmatch(device_name, changer->hdr.name, 0) == 0) {
393 const int name_len = MAX_NAME_LENGTH;
395 /* Try each device in this AutoChanger */
396 foreach_alist(device, changer->device) {
397 Dmsg1(100, "Try changer device %s\n", device->hdr.name);
399 device->dev = init_dev(jcr, NULL, device);
402 Dmsg1(100, "Device %s could not be opened. Skipped\n", dev_name.c_str());
403 Jmsg(jcr, M_WARNING, 0, _("\n"
404 " Device \"%s\" in changer \"%s\" requested by DIR could not be opened or does not exist.\n"),
405 device->hdr.name, dev_name.c_str());
408 if (!device->dev->autoselect) {
409 continue; /* device is not available */
411 dcr = new_dcr(jcr, device->dev);
413 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
414 Dmsg1(100, ">dird: %s\n", dir->msg);
419 Dmsg1(100, "Found changer device %s\n", device->hdr.name);
420 bstrncpy(dcr->pool_name, store->pool_name, name_len);
421 bstrncpy(dcr->pool_type, store->pool_type, name_len);
422 bstrncpy(dcr->media_type, store->media_type, name_len);
423 bstrncpy(dcr->dev_name, device_name, name_len);
425 if (append == SD_APPEND) {
426 ok = reserve_device_for_append(dcr);
428 ok = reserve_device_for_read(dcr);
431 Jmsg(jcr, M_WARNING, 0, _("Could not reserve device: %s\n"), dev_name.c_str());
435 Dmsg1(100, "Device %s opened.\n", device_name);
437 pm_strcpy(dev_name, device->hdr.name);
438 bash_spaces(dev_name);
439 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
440 Dmsg1(100, ">dird: %s\n", dir->msg);
443 break; /* we found it but could not open a device */
450 unbash_spaces(dir->msg);
451 pm_strcpy(jcr->errmsg, dir->msg);
452 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
454 Jmsg(jcr, M_FATAL, 0, _("\n"
455 " Device \"%s\" with MediaType \"%s\" requested by DIR not found in SD Device resources.\n"),
456 dev_name.c_str(), media_type.c_str());
457 bnet_fsend(dir, NO_device, dev_name.c_str());
458 Dmsg1(100, ">dird: %s\n", dir->msg);
461 unbash_spaces(dir->msg);
462 pm_strcpy(jcr->errmsg, dir->msg);
464 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
466 Jmsg(jcr, M_FATAL, 0, _("Bad Use Device command: %s\n"), jcr->errmsg);
467 bnet_fsend(dir, BAD_use, jcr->errmsg);
468 Dmsg1(100, ">dird: %s\n", dir->msg);
473 foreach_alist(store, dirstore) {
474 delete store->device;
483 * Query Device command from Director
484 * Sends Storage Daemon's information on the device to the
485 * caller (presumably the Director).
486 * This command always returns "true" so that the line is
487 * not closed on an error.
490 bool query_cmd(JCR *jcr)
492 POOL_MEM dev_name, VolumeName, MediaType, ChangerName;
493 BSOCK *dir = jcr->dir_bsock;
495 AUTOCHANGER *changer;
498 Dmsg1(100, "Query_cmd: %s", dir->msg);
499 ok = sscanf(dir->msg, query_device, dev_name.c_str()) == 1;
500 Dmsg1(100, "<dird: %s\n", dir->msg);
502 unbash_spaces(dev_name);
504 foreach_res(device, R_DEVICE) {
505 /* Find resource, and make sure we were able to open it */
506 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0) {
508 device->dev = init_dev(jcr, NULL, device);
514 ok = dir_update_device(jcr, device->dev);
516 ok = bnet_fsend(dir, OK_query);
518 bnet_fsend(dir, NO_query);
523 foreach_res(changer, R_AUTOCHANGER) {
524 /* Find resource, and make sure we were able to open it */
525 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
527 if (!changer->device || changer->device->size() == 0) {
528 continue; /* no devices */
530 ok = dir_update_changer(jcr, changer);
532 ok = bnet_fsend(dir, OK_query);
534 bnet_fsend(dir, NO_query);
539 /* If we get here, the device/autochanger was not found */
541 unbash_spaces(dir->msg);
542 pm_strcpy(jcr->errmsg, dir->msg);
543 bnet_fsend(dir, NO_device, dev_name.c_str());
544 Dmsg1(100, ">dird: %s\n", dir->msg);
546 unbash_spaces(dir->msg);
547 pm_strcpy(jcr->errmsg, dir->msg);
548 bnet_fsend(dir, BAD_query, jcr->errmsg);
549 Dmsg1(100, ">dird: %s\n", dir->msg);
559 * Destroy the Job Control Record and associated
560 * resources (sockets).
562 void stored_free_jcr(JCR *jcr)
564 if (jcr->file_bsock) {
565 bnet_close(jcr->file_bsock);
566 jcr->file_bsock = NULL;
569 free_pool_memory(jcr->job_name);
571 if (jcr->client_name) {
572 free_memory(jcr->client_name);
573 jcr->client_name = NULL;
575 if (jcr->fileset_name) {
576 free_memory(jcr->fileset_name);
578 if (jcr->fileset_md5) {
579 free_memory(jcr->fileset_md5);
585 if (jcr->RestoreBootstrap) {
586 unlink(jcr->RestoreBootstrap);
587 free_pool_memory(jcr->RestoreBootstrap);
588 jcr->RestoreBootstrap = NULL;
590 if (jcr->next_dev || jcr->prev_dev) {
591 Emsg0(M_FATAL, 0, _("In free_jcr(), but still attached to device!!!!\n"));
593 pthread_cond_destroy(&jcr->job_start_wait);