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\n";
58 static char NO_query[] = "3918 Query failed\n";
59 static char BAD_query[] = "3917 Bad query command: %s\n";
62 * Director requests us to start a job
63 * Basic tasks done here:
64 * - We pickup the JobId to be run from the Director.
65 * - We pickup the device, media, and pool from the Director
66 * - Wait for a connection from the File Daemon (FD)
67 * - Accept commands from the FD (i.e. run the job)
68 * - Return when the connection is terminated or
71 bool job_cmd(JCR *jcr)
75 BSOCK *dir = jcr->dir_bsock;
76 POOL_MEM job_name, client_name, job, fileset_name, fileset_md5;
77 int JobType, level, spool_attributes, no_attributes, spool_data, write_part_after_job;
81 * Get JobId and permissions from Director
83 Dmsg1(100, "<dird: %s\n", dir->msg);
84 if (sscanf(dir->msg, jobcmd, &JobId, job.c_str(), job_name.c_str(),
86 &JobType, &level, fileset_name.c_str(), &no_attributes,
87 &spool_attributes, fileset_md5.c_str(), &spool_data, &write_part_after_job) != 12) {
88 pm_strcpy(jcr->errmsg, dir->msg);
89 bnet_fsend(dir, BAD_job, jcr->errmsg);
90 Dmsg1(100, ">dird: %s\n", dir->msg);
91 Emsg1(M_FATAL, 0, _("Bad Job Command from Director: %s\n"), jcr->errmsg);
92 set_jcr_job_status(jcr, JS_ErrorTerminated);
96 * Since this job could be rescheduled, we
97 * check to see if we have it already. If so
98 * free the old jcr and use the new one.
100 ojcr = get_jcr_by_full_name(job.c_str());
101 if (ojcr && !ojcr->authenticated) {
102 Dmsg2(100, "Found ojcr=0x%x Job %s\n", (unsigned)(long)ojcr, job.c_str());
106 jcr->VolSessionId = newVolSessionId();
107 jcr->VolSessionTime = VolSessionTime;
108 bstrncpy(jcr->Job, job, sizeof(jcr->Job));
109 unbash_spaces(job_name);
110 jcr->job_name = get_pool_memory(PM_NAME);
111 pm_strcpy(jcr->job_name, job_name);
112 unbash_spaces(client_name);
113 jcr->client_name = get_pool_memory(PM_NAME);
114 pm_strcpy(jcr->client_name, client_name);
115 unbash_spaces(fileset_name);
116 jcr->fileset_name = get_pool_memory(PM_NAME);
117 pm_strcpy(jcr->fileset_name, fileset_name);
118 jcr->JobType = JobType;
119 jcr->JobLevel = level;
120 jcr->no_attributes = no_attributes;
121 jcr->spool_attributes = spool_attributes;
122 jcr->spool_data = spool_data;
123 jcr->write_part_after_job = write_part_after_job;
124 jcr->fileset_md5 = get_pool_memory(PM_NAME);
125 pm_strcpy(jcr->fileset_md5, fileset_md5);
127 jcr->authenticated = false;
130 * Pass back an authorization key for the File daemon
132 make_session_key(auth_key, NULL, 1);
133 bnet_fsend(dir, OKjob, jcr->VolSessionId, jcr->VolSessionTime, auth_key);
134 Dmsg1(100, ">dird: %s", dir->msg);
135 jcr->sd_auth_key = bstrdup(auth_key);
136 memset(auth_key, 0, sizeof(auth_key));
140 bool use_cmd(JCR *jcr)
143 * Wait for the device, media, and pool information
145 if (!use_device_cmd(jcr)) {
146 set_jcr_job_status(jcr, JS_ErrorTerminated);
147 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
153 bool run_cmd(JCR *jcr)
157 struct timespec timeout;
160 Dmsg1(100, "Run_cmd: %s\n", jcr->dir_bsock->msg);
161 /* The following jobs don't need the FD */
162 switch (jcr->JobType) {
166 jcr->authenticated = true;
171 set_jcr_job_status(jcr, JS_WaitFD); /* wait for FD to connect */
172 dir_send_job_status(jcr);
174 gettimeofday(&tv, &tz);
175 timeout.tv_nsec = tv.tv_usec * 1000;
176 timeout.tv_sec = tv.tv_sec + 30 * 60; /* wait 30 minutes */
178 Dmsg1(100, "%s waiting on FD to contact SD\n", jcr->Job);
180 * Wait for the File daemon to contact us to start the Job,
181 * when he does, we will be released, unless the 30 minutes
185 for ( ;!job_canceled(jcr); ) {
186 errstat = pthread_cond_timedwait(&jcr->job_start_wait, &jcr->mutex, &timeout);
187 if (errstat == 0 || errstat == ETIMEDOUT) {
193 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
195 if (jcr->authenticated && !job_canceled(jcr)) {
196 Dmsg1(100, "Running job %s\n", jcr->Job);
197 run_job(jcr); /* Run the job */
203 * After receiving a connection (in job.c) if it is
204 * from the File daemon, this routine is called.
206 void handle_filed_connection(BSOCK *fd, char *job_name)
210 bmicrosleep(0, 50000); /* wait 50 millisecs */
211 if (!(jcr=get_jcr_by_full_name(job_name))) {
212 Jmsg1(NULL, M_FATAL, 0, _("Job name not found: %s\n"), job_name);
213 Dmsg1(100, "Job name not found: %s\n", job_name);
217 jcr->file_bsock = fd;
218 jcr->file_bsock->jcr = jcr;
220 Dmsg1(110, "Found Job %s\n", job_name);
222 if (jcr->authenticated) {
223 Jmsg2(jcr, M_FATAL, 0, "Hey!!!! JobId %u Job %s already authenticated.\n",
224 jcr->JobId, jcr->Job);
230 * Authenticate the File daemon
232 if (jcr->authenticated || !authenticate_filed(jcr)) {
233 Dmsg1(100, "Authentication failed Job %s\n", jcr->Job);
234 Jmsg(jcr, M_FATAL, 0, _("Unable to authenticate File daemon\n"));
236 jcr->authenticated = true;
237 Dmsg1(110, "OK Authentication Job %s\n", jcr->Job);
241 if (!jcr->authenticated) {
242 set_jcr_job_status(jcr, JS_ErrorTerminated);
244 pthread_cond_signal(&jcr->job_start_wait); /* wake waiting job */
252 * Use Device command from Director
253 * He tells is what Device Name to use, the Media Type,
254 * the Pool Name, and the Pool Type.
256 * Ensure that the device exists and is opened, then store
257 * the media and pool info in the JCR.
259 static bool use_device_cmd(JCR *jcr)
261 POOL_MEM dev_name, media_type, pool_name, pool_type;
262 BSOCK *dir = jcr->dir_bsock;
265 AUTOCHANGER *changer;
270 * If there are multiple devices, the director sends us
271 * use_device for each device that it wants to use.
273 Dmsg1(100, "<dird: %s", dir->msg);
274 ok = sscanf(dir->msg, use_device, dev_name.c_str(), media_type.c_str(),
275 pool_name.c_str(), pool_type.c_str(), &PoolId, &append) == 6;
277 unbash_spaces(dev_name);
278 unbash_spaces(media_type);
279 unbash_spaces(pool_name);
280 unbash_spaces(pool_type);
282 foreach_res(device, R_DEVICE) {
283 /* Find resource, and make sure we were able to open it */
284 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0 &&
285 strcmp(device->media_type, media_type.c_str()) == 0) {
286 const int name_len = MAX_NAME_LENGTH;
290 device->dev = init_dev(jcr, NULL, device);
293 Jmsg(jcr, M_WARNING, 0, _("\n"
294 " Device \"%s\" requested by DIR could not be opened or does not exist.\n"),
296 bnet_fsend(dir, NOT_open, dev_name.c_str());
297 Dmsg1(100, ">dird: %s\n", dir->msg);
300 dcr = new_dcr(jcr, device->dev);
302 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
303 Dmsg1(100, ">dird: %s\n", dir->msg);
306 Dmsg1(100, "Found device %s\n", device->hdr.name);
307 bstrncpy(dcr->pool_name, pool_name, name_len);
308 bstrncpy(dcr->pool_type, pool_type, name_len);
309 bstrncpy(dcr->media_type, media_type, name_len);
310 bstrncpy(dcr->dev_name, dev_name, name_len);
311 dcr->PoolId = PoolId;
313 if (append == SD_APPEND) {
314 ok = reserve_device_for_append(jcr, device->dev);
316 ok = reserve_device_for_read(jcr, device->dev);
319 bnet_fsend(dir, _("3927 Could not reserve device: %s\n"), dev_name.c_str());
320 Dmsg1(100, ">dird: %s\n", dir->msg);
324 Dmsg1(220, "Got: %s", dir->msg);
325 bash_spaces(dev_name);
326 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
327 Dmsg1(100, ">dird: %s\n", dir->msg);
332 foreach_res(changer, R_AUTOCHANGER) {
333 /* Find resource, and make sure we were able to open it */
334 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
335 const int name_len = MAX_NAME_LENGTH;
337 /* Try each device in this AutoChanger */
338 foreach_alist(device, changer->device) {
339 Dmsg1(100, "Try changer device %s\n", device->hdr.name);
341 device->dev = init_dev(jcr, NULL, device);
344 Dmsg1(100, "Device %s could not be opened. Skipped\n", dev_name.c_str());
345 Jmsg(jcr, M_WARNING, 0, _("\n"
346 " Device \"%s\" in changer \"%s\" requested by DIR could not be opened or does not exist.\n"),
347 device->hdr.name, dev_name.c_str());
350 if (!device->dev->autoselect) {
351 continue; /* device is not available */
353 dcr = new_dcr(jcr, device->dev);
355 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
356 Dmsg1(100, ">dird: %s\n", dir->msg);
360 Dmsg1(100, "Found changer device %s\n", device->hdr.name);
361 bstrncpy(dcr->pool_name, pool_name, name_len);
362 bstrncpy(dcr->pool_type, pool_type, name_len);
363 bstrncpy(dcr->media_type, media_type, name_len);
364 bstrncpy(dcr->dev_name, dev_name, name_len);
366 if (append == SD_APPEND) {
367 ok = reserve_device_for_append(jcr, device->dev);
369 ok = reserve_device_for_read(jcr, device->dev);
372 Jmsg(jcr, M_WARNING, 0, _("Could not reserve device: %s\n"), dev_name.c_str());
376 Dmsg1(100, "Device %s opened.\n", dev_name.c_str());
378 pm_strcpy(dev_name, device->hdr.name);
379 bash_spaces(dev_name);
380 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
381 Dmsg1(100, ">dird: %s\n", dir->msg);
384 break; /* we found it but could not open a device */
390 unbash_spaces(dir->msg);
391 pm_strcpy(jcr->errmsg, dir->msg);
392 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
394 Jmsg(jcr, M_FATAL, 0, _("\n"
395 " Device \"%s\" with MediaType \"%s\" requested by DIR not found in SD Device resources.\n"),
396 dev_name.c_str(), media_type.c_str());
397 bnet_fsend(dir, NO_device, dev_name.c_str());
398 Dmsg1(100, ">dird: %s\n", dir->msg);
400 unbash_spaces(dir->msg);
401 pm_strcpy(jcr->errmsg, dir->msg);
403 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
405 Jmsg(jcr, M_FATAL, 0, _("Bad Use Device command: %s\n"), jcr->errmsg);
406 bnet_fsend(dir, BAD_use, jcr->errmsg);
407 Dmsg1(100, ">dird: %s\n", dir->msg);
410 return false; /* ERROR return */
414 * Query Device command from Director
415 * Sends Storage Daemon's information on the device to the
416 * caller (presumably the Director).
417 * This command always returns "true" so that the line is
418 * not closed on an error.
421 bool query_cmd(JCR *jcr)
423 POOL_MEM dev_name, VolumeName, MediaType, ChangerName;
424 BSOCK *dir = jcr->dir_bsock;
426 AUTOCHANGER *changer;
429 Dmsg1(100, "Query_cmd: %s", dir->msg);
430 ok = sscanf(dir->msg, query_device, dev_name.c_str()) == 1;
431 Dmsg1(100, "<dird: %s\n", dir->msg);
433 unbash_spaces(dev_name);
435 foreach_res(device, R_DEVICE) {
436 /* Find resource, and make sure we were able to open it */
437 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0) {
439 device->dev = init_dev(jcr, NULL, device);
445 ok = dir_update_device(jcr, device->dev);
447 ok = bnet_fsend(dir, OK_query);
449 bnet_fsend(dir, NO_query);
454 foreach_res(changer, R_AUTOCHANGER) {
455 /* Find resource, and make sure we were able to open it */
456 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
458 if (!changer->device || changer->device->size() == 0) {
459 continue; /* no devices */
461 ok = dir_update_changer(jcr, changer);
463 ok = bnet_fsend(dir, OK_query);
465 bnet_fsend(dir, NO_query);
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());
475 Dmsg1(100, ">dird: %s\n", dir->msg);
477 unbash_spaces(dir->msg);
478 pm_strcpy(jcr->errmsg, dir->msg);
479 bnet_fsend(dir, BAD_query, jcr->errmsg);
480 Dmsg1(100, ">dird: %s\n", dir->msg);
488 * Destroy the Job Control Record and associated
489 * resources (sockets).
491 void stored_free_jcr(JCR *jcr)
493 if (jcr->file_bsock) {
494 bnet_close(jcr->file_bsock);
495 jcr->file_bsock = NULL;
498 free_pool_memory(jcr->job_name);
500 if (jcr->client_name) {
501 free_memory(jcr->client_name);
502 jcr->client_name = NULL;
504 if (jcr->fileset_name) {
505 free_memory(jcr->fileset_name);
507 if (jcr->fileset_md5) {
508 free_memory(jcr->fileset_md5);
514 if (jcr->RestoreBootstrap) {
515 unlink(jcr->RestoreBootstrap);
516 free_pool_memory(jcr->RestoreBootstrap);
517 jcr->RestoreBootstrap = NULL;
519 if (jcr->next_dev || jcr->prev_dev) {
520 Emsg0(M_FATAL, 0, _("In free_jcr(), but still attached to device!!!!\n"));
522 pthread_cond_destroy(&jcr->job_start_wait);