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));
145 bool use_cmd(JCR *jcr)
148 * Wait for the device, media, and pool information
150 if (!use_storage_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.
267 char name[MAX_NAME_LENGTH];
268 char media_type[MAX_NAME_LENGTH];
269 char pool_name[MAX_NAME_LENGTH];
270 char pool_type[MAX_NAME_LENGTH];
273 static bool use_storage_cmd(JCR *jcr)
275 POOL_MEM store_name, dev_name, media_type, pool_name, pool_type;
276 BSOCK *dir = jcr->dir_bsock;
278 AUTOCHANGER *changer;
287 * If there are multiple devices, the director sends us
288 * use_device for each device that it wants to use.
290 Dmsg1(100, "<dird: %s", dir->msg);
291 dirstore = New(alist(10, not_owned_by_alist));
293 ok = sscanf(dir->msg, use_storage, store_name.c_str(),
294 media_type.c_str(), pool_name.c_str(),
295 pool_type.c_str(), &append, &Copy, &Stripe) == 7;
299 unbash_spaces(store_name);
300 unbash_spaces(media_type);
301 unbash_spaces(pool_name);
302 unbash_spaces(pool_type);
303 store = new DIRSTORE;
304 dirstore->append(store);
305 memset(store, 0, sizeof(DIRSTORE));
306 store->device = New(alist(10));
307 bstrncpy(store->name, store_name, sizeof(store->name));
308 bstrncpy(store->media_type, media_type, sizeof(store->media_type));
309 bstrncpy(store->pool_name, pool_name, sizeof(store->pool_name));
310 bstrncpy(store->pool_type, pool_type, sizeof(store->pool_type));
312 /* Now get all devices */
313 while (bnet_recv(dir) >= 0) {
314 ok = sscanf(dir->msg, use_device, dev_name.c_str()) == 1;
318 unbash_spaces(dev_name);
319 store->device->append(bstrdup(dev_name.c_str()));
321 } while (ok && bnet_recv(dir) >= 0);
323 /* This loop is debug code and can be removed */
324 /* ***FIXME*** turn off in production */
325 foreach_alist(store, dirstore) {
326 Dmsg4(100, "Storage=%s media_type=%s pool=%s pool_type=%s\n",
327 store->name, store->media_type, store->pool_name,
329 foreach_alist(device_name, store->device) {
330 Dmsg1(100, " Device=%s\n", device_name);
337 store = (DIRSTORE *)dirstore->first();
338 foreach_alist(device_name, store->device) {
339 foreach_res(device, R_DEVICE) {
340 /* Find resource, and make sure we were able to open it */
341 if (fnmatch(device_name, device->hdr.name, 0) == 0 &&
342 strcmp(device->media_type, store->media_type) == 0) {
343 const int name_len = MAX_NAME_LENGTH;
347 device->dev = init_dev(jcr, NULL, device);
350 Jmsg(jcr, M_WARNING, 0, _("\n"
351 " Device \"%s\" requested by DIR could not be opened or does not exist.\n"),
353 bnet_fsend(dir, NOT_open, dev_name.c_str());
354 Dmsg1(100, ">dird: %s\n", dir->msg);
357 dcr = new_dcr(jcr, device->dev);
359 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
360 Dmsg1(100, ">dird: %s\n", dir->msg);
363 Dmsg1(100, "Found device %s\n", device->hdr.name);
364 bstrncpy(dcr->pool_name, store->pool_name, name_len);
365 bstrncpy(dcr->pool_type, store->pool_type, name_len);
366 bstrncpy(dcr->media_type, store->media_type, name_len);
367 bstrncpy(dcr->dev_name, device_name, name_len);
369 dcr->Stripe = Stripe;
371 if (append == SD_APPEND) {
372 ok = reserve_device_for_append(dcr);
374 ok = reserve_device_for_read(dcr);
377 bnet_fsend(dir, _("3927 Could not reserve device: %s\n"), dev_name.c_str());
378 Dmsg1(100, ">dird: %s\n", dir->msg);
382 Dmsg1(220, "Got: %s", dir->msg);
383 bash_spaces(dev_name);
384 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
385 Dmsg1(100, ">dird: %s\n", dir->msg);
389 foreach_res(changer, R_AUTOCHANGER) {
390 /* Find resource, and make sure we were able to open it */
391 if (fnmatch(device_name, changer->hdr.name, 0) == 0) {
392 const int name_len = MAX_NAME_LENGTH;
394 /* Try each device in this AutoChanger */
395 foreach_alist(device, changer->device) {
396 Dmsg1(100, "Try changer device %s\n", device->hdr.name);
398 device->dev = init_dev(jcr, NULL, device);
401 Dmsg1(100, "Device %s could not be opened. Skipped\n", dev_name.c_str());
402 Jmsg(jcr, M_WARNING, 0, _("\n"
403 " Device \"%s\" in changer \"%s\" requested by DIR could not be opened or does not exist.\n"),
404 device->hdr.name, dev_name.c_str());
407 if (!device->dev->autoselect) {
408 continue; /* device is not available */
410 dcr = new_dcr(jcr, device->dev);
412 bnet_fsend(dir, _("3926 Could not get dcr for device: %s\n"), dev_name.c_str());
413 Dmsg1(100, ">dird: %s\n", dir->msg);
418 Dmsg1(100, "Found changer device %s\n", device->hdr.name);
419 bstrncpy(dcr->pool_name, store->pool_name, name_len);
420 bstrncpy(dcr->pool_type, store->pool_type, name_len);
421 bstrncpy(dcr->media_type, store->media_type, name_len);
422 bstrncpy(dcr->dev_name, device_name, name_len);
424 if (append == SD_APPEND) {
425 ok = reserve_device_for_append(dcr);
427 ok = reserve_device_for_read(dcr);
430 Jmsg(jcr, M_WARNING, 0, _("Could not reserve device: %s\n"), dev_name.c_str());
434 Dmsg1(100, "Device %s opened.\n", device_name);
436 pm_strcpy(dev_name, device->hdr.name);
437 bash_spaces(dev_name);
438 ok = bnet_fsend(dir, OK_device, dev_name.c_str());
439 Dmsg1(100, ">dird: %s\n", dir->msg);
442 break; /* we found it but could not open a device */
449 unbash_spaces(dir->msg);
450 pm_strcpy(jcr->errmsg, dir->msg);
451 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
453 Jmsg(jcr, M_FATAL, 0, _("\n"
454 " Device \"%s\" with MediaType \"%s\" requested by DIR not found in SD Device resources.\n"),
455 dev_name.c_str(), media_type.c_str());
456 bnet_fsend(dir, NO_device, dev_name.c_str());
457 Dmsg1(100, ">dird: %s\n", dir->msg);
460 unbash_spaces(dir->msg);
461 pm_strcpy(jcr->errmsg, dir->msg);
463 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
465 Jmsg(jcr, M_FATAL, 0, _("Bad Use Device command: %s\n"), jcr->errmsg);
466 bnet_fsend(dir, BAD_use, jcr->errmsg);
467 Dmsg1(100, ">dird: %s\n", dir->msg);
472 foreach_alist(store, dirstore) {
473 delete store->device;
482 * Query Device command from Director
483 * Sends Storage Daemon's information on the device to the
484 * caller (presumably the Director).
485 * This command always returns "true" so that the line is
486 * not closed on an error.
489 bool query_cmd(JCR *jcr)
491 POOL_MEM dev_name, VolumeName, MediaType, ChangerName;
492 BSOCK *dir = jcr->dir_bsock;
494 AUTOCHANGER *changer;
497 Dmsg1(100, "Query_cmd: %s", dir->msg);
498 ok = sscanf(dir->msg, query_device, dev_name.c_str()) == 1;
499 Dmsg1(100, "<dird: %s\n", dir->msg);
501 unbash_spaces(dev_name);
503 foreach_res(device, R_DEVICE) {
504 /* Find resource, and make sure we were able to open it */
505 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0) {
507 device->dev = init_dev(jcr, NULL, device);
513 ok = dir_update_device(jcr, device->dev);
515 ok = bnet_fsend(dir, OK_query);
517 bnet_fsend(dir, NO_query);
522 foreach_res(changer, R_AUTOCHANGER) {
523 /* Find resource, and make sure we were able to open it */
524 if (fnmatch(dev_name.c_str(), changer->hdr.name, 0) == 0) {
526 if (!changer->device || changer->device->size() == 0) {
527 continue; /* no devices */
529 ok = dir_update_changer(jcr, changer);
531 ok = bnet_fsend(dir, OK_query);
533 bnet_fsend(dir, NO_query);
538 /* If we get here, the device/autochanger was not found */
540 unbash_spaces(dir->msg);
541 pm_strcpy(jcr->errmsg, dir->msg);
542 bnet_fsend(dir, NO_device, dev_name.c_str());
543 Dmsg1(100, ">dird: %s\n", dir->msg);
545 unbash_spaces(dir->msg);
546 pm_strcpy(jcr->errmsg, dir->msg);
547 bnet_fsend(dir, BAD_query, jcr->errmsg);
548 Dmsg1(100, ">dird: %s\n", dir->msg);
558 * Destroy the Job Control Record and associated
559 * resources (sockets).
561 void stored_free_jcr(JCR *jcr)
563 if (jcr->file_bsock) {
564 bnet_close(jcr->file_bsock);
565 jcr->file_bsock = NULL;
568 free_pool_memory(jcr->job_name);
570 if (jcr->client_name) {
571 free_memory(jcr->client_name);
572 jcr->client_name = NULL;
574 if (jcr->fileset_name) {
575 free_memory(jcr->fileset_name);
577 if (jcr->fileset_md5) {
578 free_memory(jcr->fileset_md5);
584 if (jcr->RestoreBootstrap) {
585 unlink(jcr->RestoreBootstrap);
586 free_pool_memory(jcr->RestoreBootstrap);
587 jcr->RestoreBootstrap = NULL;
589 if (jcr->next_dev || jcr->prev_dev) {
590 Emsg0(M_FATAL, 0, _("In free_jcr(), but still attached to device!!!!\n"));
592 pthread_cond_destroy(&jcr->job_start_wait);