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_device_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";
45 static char use_device[] = "use device=%127s media_type=%127s "
46 "pool_name=%127s pool_type=%127s PoolId=%lld append=%d";
47 static char query_device[] = "query device=%127s";
50 /* Responses sent to Director daemon */
51 static char OKjob[] = "3000 OK Job SDid=%u SDtime=%u Authorization=%s\n";
52 static char OK_device[] = "3000 OK use device device=%s\n";
53 static char NO_device[] = "3924 Device \"%s\" not in SD Device resources.\n";
54 static char NOT_open[] = "3925 Device \"%s\" could not be opened or does not exist.\n";
55 static char BAD_use[] = "3913 Bad use command: %s\n";
56 static char BAD_job[] = "3915 Bad Job command: %s\n";
57 static char OK_query[] = "3001 OK query "
58 "append=%d read=%d num_writers=%d "
59 "open=%d labeled=%d offline=%d "
60 "reserved=%d max_writers=%d "
61 "autoselect=%d autochanger=%d "
63 "changer_name=%s media_type=%s volume_name=%s";
64 static char BAD_query[] = "3917 Bad query command: %s\n";
67 * Director requests us to start a job
68 * Basic tasks done here:
69 * - We pickup the JobId to be run from the Director.
70 * - We pickup the device, media, and pool from the Director
71 * - Wait for a connection from the File Daemon (FD)
72 * - Accept commands from the FD (i.e. run the job)
73 * - Return when the connection is terminated or
76 bool job_cmd(JCR *jcr)
80 BSOCK *dir = jcr->dir_bsock;
81 POOL_MEM job_name, client_name, job, fileset_name, fileset_md5;
82 int JobType, level, spool_attributes, no_attributes, spool_data, write_part_after_job;
86 * Get JobId and permissions from Director
88 Dmsg1(100, "<dird: %s\n", dir->msg);
89 if (sscanf(dir->msg, jobcmd, &JobId, job.c_str(), job_name.c_str(),
91 &JobType, &level, fileset_name.c_str(), &no_attributes,
92 &spool_attributes, fileset_md5.c_str(), &spool_data, &write_part_after_job) != 12) {
93 pm_strcpy(jcr->errmsg, dir->msg);
94 bnet_fsend(dir, BAD_job, jcr->errmsg);
95 Dmsg1(100, ">dird: %s\n", dir->msg);
96 Emsg1(M_FATAL, 0, _("Bad Job Command from Director: %s\n"), jcr->errmsg);
97 set_jcr_job_status(jcr, JS_ErrorTerminated);
101 * Since this job could be rescheduled, we
102 * check to see if we have it already. If so
103 * free the old jcr and use the new one.
105 ojcr = get_jcr_by_full_name(job.c_str());
106 if (ojcr && !ojcr->authenticated) {
107 Dmsg2(100, "Found ojcr=0x%x Job %s\n", (unsigned)(long)ojcr, job.c_str());
111 jcr->VolSessionId = newVolSessionId();
112 jcr->VolSessionTime = VolSessionTime;
113 bstrncpy(jcr->Job, job, sizeof(jcr->Job));
114 unbash_spaces(job_name);
115 jcr->job_name = get_pool_memory(PM_NAME);
116 pm_strcpy(jcr->job_name, job_name);
117 unbash_spaces(client_name);
118 jcr->client_name = get_pool_memory(PM_NAME);
119 pm_strcpy(jcr->client_name, client_name);
120 unbash_spaces(fileset_name);
121 jcr->fileset_name = get_pool_memory(PM_NAME);
122 pm_strcpy(jcr->fileset_name, fileset_name);
123 jcr->JobType = JobType;
124 jcr->JobLevel = level;
125 jcr->no_attributes = no_attributes;
126 jcr->spool_attributes = spool_attributes;
127 jcr->spool_data = spool_data;
128 jcr->write_part_after_job = write_part_after_job;
129 jcr->fileset_md5 = get_pool_memory(PM_NAME);
130 pm_strcpy(jcr->fileset_md5, fileset_md5);
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));
145 bool use_cmd(JCR *jcr)
148 * Wait for the device, media, and pool information
150 if (!use_device_cmd(jcr)) {
151 set_jcr_job_status(jcr, JS_ErrorTerminated);
152 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
158 bool run_cmd(JCR *jcr)
162 struct timespec timeout;
165 Dmsg1(100, "Run_cmd: %s\n", jcr->dir_bsock->msg);
166 /* The following jobs don't need the FD */
167 switch (jcr->JobType) {
171 jcr->authenticated = true;
176 set_jcr_job_status(jcr, JS_WaitFD); /* wait for FD to connect */
177 dir_send_job_status(jcr);
179 gettimeofday(&tv, &tz);
180 timeout.tv_nsec = tv.tv_usec * 1000;
181 timeout.tv_sec = tv.tv_sec + 30 * 60; /* wait 30 minutes */
183 Dmsg1(100, "%s waiting on FD to contact SD\n", jcr->Job);
185 * Wait for the File daemon to contact us to start the Job,
186 * when he does, we will be released, unless the 30 minutes
190 for ( ;!job_canceled(jcr); ) {
191 errstat = pthread_cond_timedwait(&jcr->job_start_wait, &jcr->mutex, &timeout);
192 if (errstat == 0 || errstat == ETIMEDOUT) {
198 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
200 if (jcr->authenticated && !job_canceled(jcr)) {
201 Dmsg1(100, "Running job %s\n", jcr->Job);
202 run_job(jcr); /* Run the job */
208 * After receiving a connection (in job.c) if it is
209 * from the File daemon, this routine is called.
211 void handle_filed_connection(BSOCK *fd, char *job_name)
215 bmicrosleep(0, 50000); /* wait 50 millisecs */
216 if (!(jcr=get_jcr_by_full_name(job_name))) {
217 Jmsg1(NULL, M_FATAL, 0, _("Job name not found: %s\n"), job_name);
218 Dmsg1(100, "Job name not found: %s\n", job_name);
222 jcr->file_bsock = fd;
223 jcr->file_bsock->jcr = jcr;
225 Dmsg1(110, "Found Job %s\n", job_name);
227 if (jcr->authenticated) {
228 Jmsg2(jcr, M_FATAL, 0, "Hey!!!! JobId %u Job %s already authenticated.\n",
229 jcr->JobId, jcr->Job);
235 * Authenticate the File daemon
237 if (jcr->authenticated || !authenticate_filed(jcr)) {
238 Dmsg1(100, "Authentication failed Job %s\n", jcr->Job);
239 Jmsg(jcr, M_FATAL, 0, _("Unable to authenticate File daemon\n"));
241 jcr->authenticated = true;
242 Dmsg1(110, "OK Authentication Job %s\n", jcr->Job);
246 if (!jcr->authenticated) {
247 set_jcr_job_status(jcr, JS_ErrorTerminated);
249 pthread_cond_signal(&jcr->job_start_wait); /* wake waiting job */
257 * Use Device command from Director
258 * He tells is what Device Name to use, the Media Type,
259 * the Pool Name, and the Pool Type.
261 * Ensure that the device exists and is opened, then store
262 * the media and pool info in the JCR.
264 static bool use_device_cmd(JCR *jcr)
266 POOL_MEM dev_name, media_type, pool_name, pool_type;
267 BSOCK *dir = jcr->dir_bsock;
270 AUTOCHANGER *changer;
275 * If there are multiple devices, the director sends us
276 * use_device for each device that it wants to use.
278 Dmsg1(100, "<dird: %s", dir->msg);
279 ok = sscanf(dir->msg, use_device, dev_name.c_str(), media_type.c_str(),
280 pool_name.c_str(), pool_type.c_str(), &PoolId, &append) == 6;
282 unbash_spaces(dev_name);
283 unbash_spaces(media_type);
284 unbash_spaces(pool_name);
285 unbash_spaces(pool_type);
287 foreach_res(device, R_DEVICE) {
288 /* Find resource, and make sure we were able to open it */
289 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0 &&
290 strcmp(device->media_type, media_type.c_str()) == 0) {
291 const int name_len = MAX_NAME_LENGTH;
295 device->dev = init_dev(jcr, NULL, device);
298 Jmsg(jcr, M_WARNING, 0, _("\n"
299 " Device \"%s\" requested by DIR could not be opened or does not exist.\n"),
301 bnet_fsend(dir, NOT_open, dev_name.c_str());
302 Dmsg1(100, ">dird: %s\n", dir->msg);
305 dcr = new_dcr(jcr, device->dev);
307 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
308 Dmsg1(100, ">dird: %s\n", dir->msg);
311 Dmsg1(100, "Found device %s\n", device->hdr.name);
312 bstrncpy(dcr->pool_name, pool_name, name_len);
313 bstrncpy(dcr->pool_type, pool_type, name_len);
314 bstrncpy(dcr->media_type, media_type, name_len);
315 bstrncpy(dcr->dev_name, dev_name, name_len);
316 dcr->PoolId = PoolId;
318 if (append == SD_APPEND) {
319 ok = reserve_device_for_append(jcr, device->dev);
321 ok = reserve_device_for_read(jcr, device->dev);
324 bnet_fsend(dir, _("3927 Could not reserve device: %s\n"), dev_name.c_str());
325 Dmsg1(100, ">dird: %s\n", dir->msg);
329 Dmsg1(220, "Got: %s", dir->msg);
330 bash_spaces(dev_name);
331 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
332 Dmsg1(100, ">dird: %s\n", dir->msg);
337 foreach_res(changer, R_AUTOCHANGER) {
338 /* Find resource, and make sure we were able to open it */
339 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
340 const int name_len = MAX_NAME_LENGTH;
342 /* Try each device in this AutoChanger */
343 foreach_alist(device, changer->device) {
344 Dmsg1(100, "Try changer device %s\n", device->hdr.name);
346 device->dev = init_dev(jcr, NULL, device);
349 Dmsg1(100, "Device %s could not be opened. Skipped\n", dev_name.c_str());
350 Jmsg(jcr, M_WARNING, 0, _("\n"
351 " Device \"%s\" in changer \"%s\" requested by DIR could not be opened or does not exist.\n"),
352 device->hdr.name, dev_name.c_str());
355 if (!device->dev->autoselect) {
356 continue; /* device is not available */
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);
365 Dmsg1(100, "Found changer device %s\n", device->hdr.name);
366 bstrncpy(dcr->pool_name, pool_name, name_len);
367 bstrncpy(dcr->pool_type, pool_type, name_len);
368 bstrncpy(dcr->media_type, media_type, name_len);
369 bstrncpy(dcr->dev_name, dev_name, name_len);
371 if (append == SD_APPEND) {
372 ok = reserve_device_for_append(jcr, device->dev);
374 ok = reserve_device_for_read(jcr, device->dev);
377 Jmsg(jcr, M_WARNING, 0, _("Could not reserve device: %s\n"), dev_name.c_str());
381 Dmsg1(100, "Device %s opened.\n", dev_name.c_str());
383 pm_strcpy(dev_name, device->hdr.name);
384 bash_spaces(dev_name);
385 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
386 Dmsg1(100, ">dird: %s\n", dir->msg);
389 break; /* we found it but could not open a device */
395 unbash_spaces(dir->msg);
396 pm_strcpy(jcr->errmsg, dir->msg);
397 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
399 Jmsg(jcr, M_FATAL, 0, _("\n"
400 " Device \"%s\" with MediaType \"%s\" requested by DIR not found in SD Device resources.\n"),
401 dev_name.c_str(), media_type.c_str());
402 bnet_fsend(dir, NO_device, dev_name.c_str());
403 Dmsg1(100, ">dird: %s\n", dir->msg);
405 unbash_spaces(dir->msg);
406 pm_strcpy(jcr->errmsg, dir->msg);
408 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
410 Jmsg(jcr, M_FATAL, 0, _("Bad Use Device command: %s\n"), jcr->errmsg);
411 bnet_fsend(dir, BAD_use, jcr->errmsg);
412 Dmsg1(100, ">dird: %s\n", dir->msg);
415 return false; /* ERROR return */
419 * Query Device command from Director
420 * Sends Storage Daemon's information on the device to the
421 * caller (presumably the Director).
422 * This command always returns "true" so that the line is
423 * not closed on an error.
426 bool query_cmd(JCR *jcr)
429 BSOCK *dir = jcr->dir_bsock;
431 AUTOCHANGER *changer;
435 Dmsg1(100, "Query_cmd: %s", dir->msg);
436 ok = sscanf(dir->msg, query_device, dev_name.c_str()) == 1;
437 Dmsg1(100, "<dird: %s\n", dir->msg);
439 unbash_spaces(dev_name);
441 foreach_res(device, R_DEVICE) {
442 /* Find resource, and make sure we were able to open it */
443 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0) {
445 device->dev = init_dev(jcr, NULL, device);
450 DEVICE *dev = device->dev;
451 POOL_MEM VolumeName, MediaType, ChangerName;
453 if (dev->is_labeled()) {
454 pm_strcpy(VolumeName, dev->VolHdr.VolName);
456 pm_strcpy(VolumeName, "*");
458 bash_spaces(VolumeName);
459 pm_strcpy(MediaType, device->media_type);
460 bash_spaces(MediaType);
461 if (device->changer_res) {
462 pm_strcpy(ChangerName, device->changer_res->hdr.name);
463 bash_spaces(ChangerName);
465 pm_strcpy(ChangerName, "*");
467 ok =bnet_fsend(dir, OK_query,
468 dev->can_append()!=0,
469 dev->can_read()!=0, dev->num_writers,
470 dev->is_open()!=0, dev->is_labeled()!=0,
471 dev->is_offline()!=0, dev->reserved_device,
472 dev->is_tape()?100000:1,
474 edit_uint64(dev->PoolId, ed1),
475 ChangerName.c_str(), MediaType.c_str(), VolumeName.c_str());
476 Dmsg1(100, ">dird: %s\n", dir->msg);
480 foreach_res(changer, R_AUTOCHANGER) {
481 /* Find resource, and make sure we were able to open it */
482 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
484 /* This is mostly to indicate that we are here */
485 ok = bnet_fsend(dir, OK_query,
486 0, 0, 0, /* append, read, num_writers */
487 0, 0, 0, /* is_open, is_labeled, offline */
488 0, 0, /* reserved, max_writers */
489 0, 1, /* autoselect, AutoChanger = 1 */
491 "*", "*", "*"); /* ChangerName, MediaType, VolName */
492 Dmsg1(100, ">dird: %s\n", dir->msg);
496 /* If we get here, the device/autochanger was not found */
498 unbash_spaces(dir->msg);
499 pm_strcpy(jcr->errmsg, dir->msg);
500 bnet_fsend(dir, NO_device, dev_name.c_str());
501 Dmsg1(100, ">dird: %s\n", dir->msg);
503 unbash_spaces(dir->msg);
504 pm_strcpy(jcr->errmsg, dir->msg);
505 bnet_fsend(dir, BAD_query, jcr->errmsg);
506 Dmsg1(100, ">dird: %s\n", dir->msg);
514 * Destroy the Job Control Record and associated
515 * resources (sockets).
517 void stored_free_jcr(JCR *jcr)
519 if (jcr->file_bsock) {
520 bnet_close(jcr->file_bsock);
521 jcr->file_bsock = NULL;
524 free_pool_memory(jcr->job_name);
526 if (jcr->client_name) {
527 free_memory(jcr->client_name);
528 jcr->client_name = NULL;
530 if (jcr->fileset_name) {
531 free_memory(jcr->fileset_name);
533 if (jcr->fileset_md5) {
534 free_memory(jcr->fileset_md5);
540 if (jcr->RestoreBootstrap) {
541 unlink(jcr->RestoreBootstrap);
542 free_pool_memory(jcr->RestoreBootstrap);
543 jcr->RestoreBootstrap = NULL;
545 if (jcr->next_dev || jcr->prev_dev) {
546 Emsg0(M_FATAL, 0, _("In free_jcr(), but still attached to device!!!!\n"));
548 pthread_cond_destroy(&jcr->job_start_wait);