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 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 append=%d read=%d num_writers=%d "
58 "num_waiting=%d open=%d use_count=%d labeled=%d offline=%d "
59 "autoselect=%d autochanger=%d "
60 "changer_name=%s media_type=%s volume_name=%s";
61 static char BAD_query[] = "3917 Bad query command: %s\n";
64 * Director requests us to start a job
65 * Basic tasks done here:
66 * - We pickup the JobId to be run from the Director.
67 * - We pickup the device, media, and pool from the Director
68 * - Wait for a connection from the File Daemon (FD)
69 * - Accept commands from the FD (i.e. run the job)
70 * - Return when the connection is terminated or
73 bool job_cmd(JCR *jcr)
77 BSOCK *dir = jcr->dir_bsock;
78 POOL_MEM job_name, client_name, job, fileset_name, fileset_md5;
79 int JobType, level, spool_attributes, no_attributes, spool_data, write_part_after_job;
83 * Get JobId and permissions from Director
85 Dmsg1(100, "Job_cmd: %s\n", dir->msg);
86 if (sscanf(dir->msg, jobcmd, &JobId, job.c_str(), job_name.c_str(),
88 &JobType, &level, fileset_name.c_str(), &no_attributes,
89 &spool_attributes, fileset_md5.c_str(), &spool_data, &write_part_after_job) != 12) {
90 pm_strcpy(jcr->errmsg, dir->msg);
91 bnet_fsend(dir, BAD_job, jcr->errmsg);
92 Emsg1(M_FATAL, 0, _("Bad Job Command from Director: %s\n"), jcr->errmsg);
93 set_jcr_job_status(jcr, JS_ErrorTerminated);
97 * Since this job could be rescheduled, we
98 * check to see if we have it already. If so
99 * free the old jcr and use the new one.
101 ojcr = get_jcr_by_full_name(job.c_str());
102 if (ojcr && !ojcr->authenticated) {
103 Dmsg2(100, "Found ojcr=0x%x Job %s\n", (unsigned)(long)ojcr, job.c_str());
107 jcr->VolSessionId = newVolSessionId();
108 jcr->VolSessionTime = VolSessionTime;
109 bstrncpy(jcr->Job, job, sizeof(jcr->Job));
110 unbash_spaces(job_name);
111 jcr->job_name = get_pool_memory(PM_NAME);
112 pm_strcpy(jcr->job_name, job_name);
113 unbash_spaces(client_name);
114 jcr->client_name = get_pool_memory(PM_NAME);
115 pm_strcpy(jcr->client_name, client_name);
116 unbash_spaces(fileset_name);
117 jcr->fileset_name = get_pool_memory(PM_NAME);
118 pm_strcpy(jcr->fileset_name, fileset_name);
119 jcr->JobType = JobType;
120 jcr->JobLevel = level;
121 jcr->no_attributes = no_attributes;
122 jcr->spool_attributes = spool_attributes;
123 jcr->spool_data = spool_data;
124 jcr->write_part_after_job = write_part_after_job;
125 jcr->fileset_md5 = get_pool_memory(PM_NAME);
126 pm_strcpy(jcr->fileset_md5, fileset_md5);
128 jcr->authenticated = false;
131 * Pass back an authorization key for the File daemon
133 make_session_key(auth_key, NULL, 1);
134 bnet_fsend(dir, OKjob, jcr->VolSessionId, jcr->VolSessionTime, auth_key);
135 Dmsg1(110, ">dird: %s", dir->msg);
136 jcr->sd_auth_key = bstrdup(auth_key);
137 memset(auth_key, 0, sizeof(auth_key));
141 bool use_cmd(JCR *jcr)
144 * Wait for the device, media, and pool information
146 Dmsg1(100, "Use_cmd: %s\n", jcr->dir_bsock->msg);
147 if (!use_device_cmd(jcr)) {
148 set_jcr_job_status(jcr, JS_ErrorTerminated);
149 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
155 bool run_cmd(JCR *jcr)
159 struct timespec timeout;
162 Dmsg1(100, "Run_cmd: %s\n", jcr->dir_bsock->msg);
163 /* The following jobs don't need the FD */
164 switch (jcr->JobType) {
168 jcr->authenticated = true;
173 set_jcr_job_status(jcr, JS_WaitFD); /* wait for FD to connect */
174 dir_send_job_status(jcr);
176 gettimeofday(&tv, &tz);
177 timeout.tv_nsec = tv.tv_usec * 1000;
178 timeout.tv_sec = tv.tv_sec + 30 * 60; /* wait 30 minutes */
180 Dmsg1(100, "%s waiting on FD to contact SD\n", jcr->Job);
182 * Wait for the File daemon to contact us to start the Job,
183 * when he does, we will be released, unless the 30 minutes
187 for ( ;!job_canceled(jcr); ) {
188 errstat = pthread_cond_timedwait(&jcr->job_start_wait, &jcr->mutex, &timeout);
189 if (errstat == 0 || errstat == ETIMEDOUT) {
195 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
197 if (jcr->authenticated && !job_canceled(jcr)) {
198 Dmsg1(100, "Running job %s\n", jcr->Job);
199 run_job(jcr); /* Run the job */
205 * After receiving a connection (in job.c) if it is
206 * from the File daemon, this routine is called.
208 void handle_filed_connection(BSOCK *fd, char *job_name)
212 bmicrosleep(0, 50000); /* wait 50 millisecs */
213 if (!(jcr=get_jcr_by_full_name(job_name))) {
214 Jmsg1(NULL, M_FATAL, 0, _("Job name not found: %s\n"), job_name);
215 Dmsg1(100, "Job name not found: %s\n", job_name);
219 jcr->file_bsock = fd;
220 jcr->file_bsock->jcr = jcr;
222 Dmsg1(110, "Found Job %s\n", job_name);
224 if (jcr->authenticated) {
225 Jmsg2(jcr, M_FATAL, 0, "Hey!!!! JobId %u Job %s already authenticated.\n",
226 jcr->JobId, jcr->Job);
232 * Authenticate the File daemon
234 if (jcr->authenticated || !authenticate_filed(jcr)) {
235 Dmsg1(100, "Authentication failed Job %s\n", jcr->Job);
236 Jmsg(jcr, M_FATAL, 0, _("Unable to authenticate File daemon\n"));
238 jcr->authenticated = true;
239 Dmsg1(110, "OK Authentication Job %s\n", jcr->Job);
243 if (!jcr->authenticated) {
244 set_jcr_job_status(jcr, JS_ErrorTerminated);
246 pthread_cond_signal(&jcr->job_start_wait); /* wake waiting job */
254 * Use Device command from Director
255 * He tells is what Device Name to use, the Media Type,
256 * the Pool Name, and the Pool Type.
258 * Ensure that the device exists and is opened, then store
259 * the media and pool info in the JCR.
261 static bool use_device_cmd(JCR *jcr)
263 POOL_MEM dev_name, media_type, pool_name, pool_type;
264 BSOCK *dir = jcr->dir_bsock;
266 AUTOCHANGER *changer;
270 Dmsg1(100, "Use_device_cmd: %s", jcr->dir_bsock->msg);
272 * If there are multiple devices, the director sends us
273 * use_device for each device that it wants to use.
275 ok = sscanf(dir->msg, use_device, dev_name.c_str(), media_type.c_str(),
276 pool_name.c_str(), pool_type.c_str(), &append) == 5;
278 unbash_spaces(dev_name);
279 unbash_spaces(media_type);
280 unbash_spaces(pool_name);
281 unbash_spaces(pool_type);
283 foreach_res(device, R_DEVICE) {
284 /* Find resource, and make sure we were able to open it */
285 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0 &&
286 strcmp(device->media_type, media_type.c_str()) == 0) {
287 const int name_len = MAX_NAME_LENGTH;
291 device->dev = init_dev(jcr, NULL, device);
294 Jmsg(jcr, M_WARNING, 0, _("\n"
295 " Device \"%s\" requested by DIR could not be opened or does not exist.\n"),
297 bnet_fsend(dir, NOT_open, dev_name.c_str());
300 dcr = new_dcr(jcr, device->dev);
302 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
305 Dmsg1(100, "Found device %s\n", device->hdr.name);
306 bstrncpy(dcr->pool_name, pool_name, name_len);
307 bstrncpy(dcr->pool_type, pool_type, name_len);
308 bstrncpy(dcr->media_type, media_type, name_len);
309 bstrncpy(dcr->dev_name, dev_name, name_len);
311 if (append == SD_APPEND) {
312 ok = reserve_device_for_append(jcr, device->dev);
314 ok = reserve_device_for_read(jcr, device->dev);
317 bnet_fsend(dir, _("3927 Could not reserve device: %s\n"), dev_name.c_str());
321 Dmsg1(220, "Got: %s", dir->msg);
322 bash_spaces(dev_name);
323 return bnet_fsend(dir, OK_device, dev_name.c_str());
327 foreach_res(changer, R_AUTOCHANGER) {
328 /* Find resource, and make sure we were able to open it */
329 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
330 const int name_len = MAX_NAME_LENGTH;
332 /* Try each device in this AutoChanger */
333 foreach_alist(device, changer->device) {
334 Dmsg1(100, "Try changer device %s\n", device->hdr.name);
336 device->dev = init_dev(jcr, NULL, device);
339 Dmsg1(100, "Device %s could not be opened. Skipped\n", dev_name.c_str());
340 Jmsg(jcr, M_WARNING, 0, _("\n"
341 " Device \"%s\" in changer \"%s\" requested by DIR could not be opened or does not exist.\n"),
342 device->hdr.name, dev_name.c_str());
345 if (!device->dev->autoselect) {
346 continue; /* device is not available */
348 dcr = new_dcr(jcr, device->dev);
350 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
354 Dmsg1(100, "Found changer device %s\n", device->hdr.name);
355 bstrncpy(dcr->pool_name, pool_name, name_len);
356 bstrncpy(dcr->pool_type, pool_type, name_len);
357 bstrncpy(dcr->media_type, media_type, name_len);
358 bstrncpy(dcr->dev_name, dev_name, name_len);
360 if (append == SD_APPEND) {
361 ok = reserve_device_for_append(jcr, device->dev);
363 ok = reserve_device_for_read(jcr, device->dev);
366 Jmsg(jcr, M_WARNING, 0, _("Could not reserve device: %s\n"), dev_name.c_str());
370 Dmsg1(100, "Device %s opened.\n", dev_name.c_str());
372 pm_strcpy(dev_name, device->hdr.name);
373 bash_spaces(dev_name);
374 return bnet_fsend(dir, OK_device, dev_name.c_str());
376 break; /* we found it but could not open a device */
382 unbash_spaces(dir->msg);
383 pm_strcpy(jcr->errmsg, dir->msg);
384 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
386 Jmsg(jcr, M_FATAL, 0, _("\n"
387 " Device \"%s\" with MediaType \"%s\" requested by DIR not found in SD Device resources.\n"),
388 dev_name.c_str(), media_type.c_str());
389 bnet_fsend(dir, NO_device, dev_name.c_str());
391 unbash_spaces(dir->msg);
392 pm_strcpy(jcr->errmsg, dir->msg);
394 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
396 Jmsg(jcr, M_FATAL, 0, _("Bad Use Device command: %s\n"), jcr->errmsg);
397 bnet_fsend(dir, BAD_use, jcr->errmsg);
400 return false; /* ERROR return */
404 * Query Device command from Director
405 * Sends Storage Daemon's information on the device to the
406 * caller (presumably the Director).
407 * This command always returns "true" so that the line is
408 * not closed on an error.
411 bool query_cmd(JCR *jcr)
414 BSOCK *dir = jcr->dir_bsock;
416 AUTOCHANGER *changer;
419 Dmsg1(100, "Query_cmd: %s", dir->msg);
421 ok = sscanf(dir->msg, query_device, dev_name.c_str()) == 1;
423 unbash_spaces(dev_name);
425 foreach_res(device, R_DEVICE) {
426 /* Find resource, and make sure we were able to open it */
427 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0) {
429 device->dev = init_dev(jcr, NULL, device);
434 DEVICE *dev = device->dev;
435 POOL_MEM VolumeName, MediaType, ChangerName;
437 if (dev->is_labeled()) {
438 pm_strcpy(VolumeName, dev->VolHdr.VolName);
440 pm_strcpy(VolumeName, "");
442 bash_spaces(VolumeName);
443 pm_strcpy(MediaType, device->media_type);
444 bash_spaces(MediaType);
445 if (device->changer_res) {
446 pm_strcpy(ChangerName, device->changer_res->hdr.name);
447 bash_spaces(ChangerName);
449 pm_strcpy(ChangerName, "");
451 return bnet_fsend(dir, OK_query, dev->can_append()!=0,
452 dev->can_read()!=0, dev->num_writers, dev->num_waiting,
453 dev->is_open()!=0, dev->use_count, dev->is_labeled()!=0,
454 dev->is_offline()!=0, 0, dev->autoselect,
455 ChangerName.c_str(), MediaType.c_str(), VolumeName.c_str());
458 foreach_res(changer, R_AUTOCHANGER) {
459 /* Find resource, and make sure we were able to open it */
460 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
462 /* This is mostly to indicate that we are here */
463 return bnet_fsend(dir, OK_query, 0,
466 0, 1, "*", /* Set AutoChanger = 1 */
470 /* If we get here, the device/autochanger was not found */
472 unbash_spaces(dir->msg);
473 pm_strcpy(jcr->errmsg, dir->msg);
474 bnet_fsend(dir, NO_device, dev_name.c_str());
476 unbash_spaces(dir->msg);
477 pm_strcpy(jcr->errmsg, dir->msg);
478 bnet_fsend(dir, BAD_query, jcr->errmsg);
486 * Destroy the Job Control Record and associated
487 * resources (sockets).
489 void stored_free_jcr(JCR *jcr)
491 if (jcr->file_bsock) {
492 bnet_close(jcr->file_bsock);
493 jcr->file_bsock = NULL;
496 free_pool_memory(jcr->job_name);
498 if (jcr->client_name) {
499 free_memory(jcr->client_name);
500 jcr->client_name = NULL;
502 if (jcr->fileset_name) {
503 free_memory(jcr->fileset_name);
505 if (jcr->fileset_md5) {
506 free_memory(jcr->fileset_md5);
512 if (jcr->RestoreBootstrap) {
513 unlink(jcr->RestoreBootstrap);
514 free_pool_memory(jcr->RestoreBootstrap);
515 jcr->RestoreBootstrap = NULL;
517 if (jcr->next_dev || jcr->prev_dev) {
518 Emsg0(M_FATAL, 0, _("In free_jcr(), but still attached to device!!!!\n"));
520 pthread_cond_destroy(&jcr->job_start_wait);