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\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 "
59 "media_type=%s volume_name=%s";
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, write_part_after_job;
82 * Get JobId and permissions from Director
84 Dmsg1(100, "Job_cmd: %s\n", dir->msg);
85 if (sscanf(dir->msg, jobcmd, &JobId, job.c_str(), job_name.c_str(),
87 &JobType, &level, fileset_name.c_str(), &no_attributes,
88 &spool_attributes, fileset_md5.c_str(), &spool_data, &write_part_after_job) != 12) {
89 pm_strcpy(jcr->errmsg, dir->msg);
90 bnet_fsend(dir, BAD_job, jcr->errmsg);
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(110, ">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 /* The following jobs don't need the FD */
161 switch (jcr->JobType) {
165 jcr->authenticated = true;
170 set_jcr_job_status(jcr, JS_WaitFD); /* wait for FD to connect */
171 dir_send_job_status(jcr);
173 gettimeofday(&tv, &tz);
174 timeout.tv_nsec = tv.tv_usec * 1000;
175 timeout.tv_sec = tv.tv_sec + 30 * 60; /* wait 30 minutes */
177 Dmsg1(100, "%s waiting on FD to contact SD\n", jcr->Job);
179 * Wait for the File daemon to contact us to start the Job,
180 * when he does, we will be released, unless the 30 minutes
184 for ( ;!job_canceled(jcr); ) {
185 errstat = pthread_cond_timedwait(&jcr->job_start_wait, &jcr->mutex, &timeout);
186 if (errstat == 0 || errstat == ETIMEDOUT) {
192 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
194 if (jcr->authenticated && !job_canceled(jcr)) {
195 Dmsg1(100, "Running job %s\n", jcr->Job);
196 run_job(jcr); /* Run the job */
202 * After receiving a connection (in job.c) if it is
203 * from the File daemon, this routine is called.
205 void handle_filed_connection(BSOCK *fd, char *job_name)
209 bmicrosleep(0, 50000); /* wait 50 millisecs */
210 if (!(jcr=get_jcr_by_full_name(job_name))) {
211 Jmsg1(NULL, M_FATAL, 0, _("Job name not found: %s\n"), job_name);
212 Dmsg1(100, "Job name not found: %s\n", job_name);
216 jcr->file_bsock = fd;
217 jcr->file_bsock->jcr = jcr;
219 Dmsg1(110, "Found Job %s\n", job_name);
221 if (jcr->authenticated) {
222 Jmsg2(jcr, M_FATAL, 0, "Hey!!!! JobId %u Job %s already authenticated.\n",
223 jcr->JobId, jcr->Job);
229 * Authenticate the File daemon
231 if (jcr->authenticated || !authenticate_filed(jcr)) {
232 Dmsg1(100, "Authentication failed Job %s\n", jcr->Job);
233 Jmsg(jcr, M_FATAL, 0, _("Unable to authenticate File daemon\n"));
235 jcr->authenticated = true;
236 Dmsg1(110, "OK Authentication Job %s\n", jcr->Job);
240 if (!jcr->authenticated) {
241 set_jcr_job_status(jcr, JS_ErrorTerminated);
243 pthread_cond_signal(&jcr->job_start_wait); /* wake waiting job */
251 * Use Device command from Director
252 * He tells is what Device Name to use, the Media Type,
253 * the Pool Name, and the Pool Type.
255 * Ensure that the device exists and is opened, then store
256 * the media and pool info in the JCR.
258 static bool use_device_cmd(JCR *jcr)
260 POOL_MEM dev_name, media_type, pool_name, pool_type;
261 BSOCK *dir = jcr->dir_bsock;
266 Dmsg1(120, "Use device: %s", dir->msg);
268 * If there are multiple devices, the director sends us
269 * use_device for each device that it wants to use.
271 ok = sscanf(dir->msg, use_device, dev_name.c_str(), media_type.c_str(),
272 pool_name.c_str(), pool_type.c_str(), &append) == 5;
274 unbash_spaces(dev_name);
275 unbash_spaces(media_type);
276 unbash_spaces(pool_name);
277 unbash_spaces(pool_type);
279 foreach_res(device, R_DEVICE) {
280 /* Find resource, and make sure we were able to open it */
281 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0 &&
282 strcmp(device->media_type, media_type.c_str()) == 0) {
283 const int name_len = MAX_NAME_LENGTH;
287 device->dev = init_dev(jcr, NULL, device);
290 Jmsg(jcr, M_FATAL, 0, _("\n"
291 " Archive \"%s\" requested by DIR could not be opened or does not exist.\n"),
293 bnet_fsend(dir, NOT_open, dev_name.c_str());
296 dcr = new_dcr(jcr, device->dev);
298 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
301 Dmsg1(120, "Found device %s\n", device->hdr.name);
302 bstrncpy(dcr->pool_name, pool_name, name_len);
303 bstrncpy(dcr->pool_type, pool_type, name_len);
304 bstrncpy(dcr->media_type, media_type, name_len);
305 bstrncpy(dcr->dev_name, dev_name, name_len);
307 if (append == SD_APPEND) {
308 ok = reserve_device_for_append(jcr, device->dev);
310 ok = reserve_device_for_read(jcr, device->dev);
313 bnet_fsend(dir, _("3927 Could not reserve device: %s\n"), dev_name.c_str());
317 Dmsg1(220, "Got: %s", dir->msg);
318 return bnet_fsend(dir, OK_device);
323 unbash_spaces(dir->msg);
324 pm_strcpy(jcr->errmsg, dir->msg);
325 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
327 Jmsg(jcr, M_FATAL, 0, _("\n"
328 " Device \"%s\" with MediaType \"%s\" requested by DIR not found in SD Device resources.\n"),
329 dev_name.c_str(), media_type.c_str());
330 bnet_fsend(dir, NO_device, dev_name.c_str());
332 unbash_spaces(dir->msg);
333 pm_strcpy(jcr->errmsg, dir->msg);
335 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
337 Jmsg(jcr, M_FATAL, 0, _("Bad Use Device command: %s\n"), jcr->errmsg);
338 bnet_fsend(dir, BAD_use, jcr->errmsg);
341 return false; /* ERROR return */
345 * Query Device command from Director
346 * Sends Storage Daemon's information on the device to the
347 * caller (presumably the Director).
348 * This command always returns "true" so that the line is
349 * not closed on an error.
352 bool query_cmd(JCR *jcr)
355 BSOCK *dir = jcr->dir_bsock;
359 Dmsg1(120, "Query: %s", dir->msg);
361 ok = sscanf(dir->msg, query_device, dev_name.c_str()) == 1;
363 unbash_spaces(dev_name);
365 foreach_res(device, R_DEVICE) {
366 /* Find resource, and make sure we were able to open it */
367 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0) {
369 device->dev = init_dev(jcr, NULL, device);
374 DEVICE *dev = device->dev;
375 POOL_MEM VolumeName, MediaType;
377 if (dev->is_labeled()) {
378 pm_strcpy(VolumeName, dev->VolHdr.VolName);
380 pm_strcpy(VolumeName, "");
382 bash_spaces(VolumeName);
383 pm_strcpy(MediaType, device->media_type);
384 bash_spaces(MediaType);
385 return bnet_fsend(dir, OK_query, dev->can_append()!=0,
386 dev->can_read()!=0, dev->num_writers, dev->num_waiting,
387 dev->is_open()!=0, dev->use_count, dev->is_labeled()!=0,
388 MediaType.c_str(), VolumeName.c_str());
392 unbash_spaces(dir->msg);
393 pm_strcpy(jcr->errmsg, dir->msg);
394 bnet_fsend(dir, NO_device, dev_name.c_str());
396 unbash_spaces(dir->msg);
397 pm_strcpy(jcr->errmsg, dir->msg);
398 bnet_fsend(dir, BAD_query, jcr->errmsg);
405 * Destroy the Job Control Record and associated
406 * resources (sockets).
408 void stored_free_jcr(JCR *jcr)
410 if (jcr->file_bsock) {
411 bnet_close(jcr->file_bsock);
412 jcr->file_bsock = NULL;
415 free_pool_memory(jcr->job_name);
417 if (jcr->client_name) {
418 free_memory(jcr->client_name);
419 jcr->client_name = NULL;
421 if (jcr->fileset_name) {
422 free_memory(jcr->fileset_name);
424 if (jcr->fileset_md5) {
425 free_memory(jcr->fileset_md5);
431 if (jcr->RestoreBootstrap) {
432 unlink(jcr->RestoreBootstrap);
433 free_pool_memory(jcr->RestoreBootstrap);
434 jcr->RestoreBootstrap = NULL;
436 if (jcr->next_dev || jcr->prev_dev) {
437 Emsg0(M_FATAL, 0, _("In free_jcr(), but still attached to device!!!!\n"));
439 pthread_cond_destroy(&jcr->job_start_wait);