2 * Job control and execution for Storage Daemon
10 Copyright (C) 2000-2004 Kern Sibbald and John Walker
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 "
45 static char use_device[] = "use device=%127s media_type=%127s pool_name=%127s pool_type=%127s\n";
47 /* Responses sent to Director daemon */
48 static char OKjob[] = "3000 OK Job SDid=%u SDtime=%u Authorization=%s\n";
49 static char OK_device[] = "3000 OK use device\n";
50 static char NO_device[] = "3914 Device \"%s\" not in SD Device resources.\n";
51 static char BAD_use[] = "3913 Bad use command: %s\n";
52 static char BAD_job[] = "3915 Bad Job command: %s\n";
55 * Director requests us to start a job
56 * Basic tasks done here:
57 * - We pickup the JobId to be run from the Director.
58 * - We pickup the device, media, and pool from the Director
59 * - Wait for a connection from the File Daemon (FD)
60 * - Accept commands from the FD (i.e. run the job)
61 * - Return when the connection is terminated or
64 bool job_cmd(JCR *jcr)
68 BSOCK *dir = jcr->dir_bsock;
69 POOL_MEM job_name, client_name, job, fileset_name, fileset_md5;
70 int JobType, level, spool_attributes, no_attributes, spool_data;
73 struct timespec timeout;
77 * Get JobId and permissions from Director
79 Dmsg1(100, "Job_cmd: %s\n", dir->msg);
80 if (sscanf(dir->msg, jobcmd, &JobId, job.c_str(), job_name.c_str(),
82 &JobType, &level, fileset_name.c_str(), &no_attributes,
83 &spool_attributes, fileset_md5.c_str(), &spool_data) != 11) {
84 pm_strcpy(jcr->errmsg, dir->msg);
85 bnet_fsend(dir, BAD_job, jcr->errmsg);
86 Emsg1(M_FATAL, 0, _("Bad Job Command from Director: %s\n"), jcr->errmsg);
87 set_jcr_job_status(jcr, JS_ErrorTerminated);
91 * Since this job could be rescheduled, we
92 * check to see if we have it already. If so
93 * free the old jcr and use the new one.
95 ojcr = get_jcr_by_full_name(job.c_str());
96 if (ojcr && !ojcr->authenticated) {
97 Dmsg2(100, "Found ojcr=0x%x Job %s\n", (unsigned)(long)ojcr, job.c_str());
101 jcr->VolSessionId = newVolSessionId();
102 jcr->VolSessionTime = VolSessionTime;
103 bstrncpy(jcr->Job, job, sizeof(jcr->Job));
104 unbash_spaces(job_name);
105 jcr->job_name = get_pool_memory(PM_NAME);
106 pm_strcpy(jcr->job_name, job_name);
107 unbash_spaces(client_name);
108 jcr->client_name = get_pool_memory(PM_NAME);
109 pm_strcpy(jcr->client_name, client_name);
110 unbash_spaces(fileset_name);
111 jcr->fileset_name = get_pool_memory(PM_NAME);
112 pm_strcpy(jcr->fileset_name, fileset_name);
113 jcr->JobType = JobType;
114 jcr->JobLevel = level;
115 jcr->no_attributes = no_attributes;
116 jcr->spool_attributes = spool_attributes;
117 jcr->spool_data = spool_data;
118 jcr->fileset_md5 = get_pool_memory(PM_NAME);
119 pm_strcpy(jcr->fileset_md5, fileset_md5);
121 jcr->authenticated = false;
124 * Pass back an authorization key for the File daemon
126 make_session_key(auth_key, NULL, 1);
127 bnet_fsend(dir, OKjob, jcr->VolSessionId, jcr->VolSessionTime, auth_key);
128 Dmsg1(110, ">dird: %s", dir->msg);
129 jcr->sd_auth_key = bstrdup(auth_key);
130 memset(auth_key, 0, sizeof(auth_key));
133 * Wait for the device, media, and pool information
135 if (!use_device_cmd(jcr)) {
136 set_jcr_job_status(jcr, JS_ErrorTerminated);
137 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
141 /* The following jobs don't need the FD */
142 switch (jcr->JobType) {
146 jcr->authenticated = true;
151 set_jcr_job_status(jcr, JS_WaitFD); /* wait for FD to connect */
152 dir_send_job_status(jcr);
154 gettimeofday(&tv, &tz);
155 timeout.tv_nsec = tv.tv_usec * 1000;
156 timeout.tv_sec = tv.tv_sec + 30 * 60; /* wait 30 minutes */
158 Dmsg1(100, "%s waiting on FD to contact SD\n", jcr->Job);
160 * Wait for the File daemon to contact us to start the Job,
161 * when he does, we will be released, unless the 30 minutes
165 for ( ;!job_canceled(jcr); ) {
166 errstat = pthread_cond_timedwait(&jcr->job_start_wait, &jcr->mutex, &timeout);
167 if (errstat == 0 || errstat == ETIMEDOUT) {
173 memset(jcr->sd_auth_key, 0, strlen(jcr->sd_auth_key));
175 if (jcr->authenticated && !job_canceled(jcr)) {
176 Dmsg1(100, "Running job %s\n", jcr->Job);
177 run_job(jcr); /* Run the job */
183 * After receiving a connection (in job.c) if it is
184 * from the File daemon, this routine is called.
186 void handle_filed_connection(BSOCK *fd, char *job_name)
190 bmicrosleep(0, 50000); /* wait 50 millisecs */
191 if (!(jcr=get_jcr_by_full_name(job_name))) {
192 Jmsg1(NULL, M_FATAL, 0, _("Job name not found: %s\n"), job_name);
193 Dmsg1(100, "Job name not found: %s\n", job_name);
197 jcr->file_bsock = fd;
198 jcr->file_bsock->jcr = jcr;
200 Dmsg1(110, "Found Job %s\n", job_name);
202 if (jcr->authenticated) {
203 Jmsg2(jcr, M_FATAL, 0, "Hey!!!! JobId %u Job %s already authenticated.\n",
204 jcr->JobId, jcr->Job);
210 * Authenticate the File daemon
212 if (jcr->authenticated || !authenticate_filed(jcr)) {
213 Dmsg1(100, "Authentication failed Job %s\n", jcr->Job);
214 Jmsg(jcr, M_FATAL, 0, _("Unable to authenticate File daemon\n"));
216 jcr->authenticated = true;
217 Dmsg1(110, "OK Authentication Job %s\n", jcr->Job);
221 if (!jcr->authenticated) {
222 set_jcr_job_status(jcr, JS_ErrorTerminated);
224 pthread_cond_signal(&jcr->job_start_wait); /* wake waiting job */
232 * Use Device command from Director
233 * He tells is what Device Name to use, the Media Type,
234 * the Pool Name, and the Pool Type.
236 * Ensure that the device exists and is opened, then store
237 * the media and pool info in the JCR.
239 static bool use_device_cmd(JCR *jcr)
241 POOL_MEM dev_name, media_type, pool_name, pool_type;
242 BSOCK *dir = jcr->dir_bsock;
245 if (bnet_recv(dir) <= 0) {
246 Jmsg0(jcr, M_FATAL, 0, _("No Device from Director\n"));
250 Dmsg1(120, "Use device: %s", dir->msg);
251 if (sscanf(dir->msg, use_device, dev_name.c_str(), media_type.c_str(),
252 pool_name.c_str(), pool_type.c_str()) == 4) {
253 unbash_spaces(dev_name);
254 unbash_spaces(media_type);
255 unbash_spaces(pool_name);
256 unbash_spaces(pool_type);
258 foreach_res(device, R_DEVICE) {
259 /* Find resource, and make sure we were able to open it */
260 if (fnmatch(dev_name.c_str(), device->hdr.name, 0) == 0 &&
261 device->dev && strcmp(device->media_type, media_type.c_str()) == 0) {
262 const int name_len = MAX_NAME_LENGTH;
265 dcr = new_dcr(jcr, device->dev);
269 Dmsg1(120, "Found device %s\n", device->hdr.name);
270 bstrncpy(dcr->pool_name, pool_name, name_len);
271 bstrncpy(dcr->pool_type, pool_type, name_len);
272 bstrncpy(dcr->media_type, media_type, name_len);
273 bstrncpy(dcr->dev_name, dev_name, name_len);
274 jcr->device = device;
275 Dmsg4(120, use_device, dev_name.c_str(), media_type.c_str(), pool_name.c_str(), pool_type.c_str());
276 return bnet_fsend(dir, OK_device);
281 unbash_spaces(dir->msg);
282 pm_strcpy(jcr->errmsg, dir->msg);
283 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
285 Jmsg(jcr, M_FATAL, 0, _("\n"
286 " Device \"%s\" with MediaType \"%s\" requested by Dir not found in SD Device resources.\n"),
287 dev_name.c_str(), media_type.c_str());
288 bnet_fsend(dir, NO_device, dev_name.c_str());
290 unbash_spaces(dir->msg);
291 pm_strcpy(jcr->errmsg, dir->msg);
293 Jmsg(jcr, M_INFO, 0, _("Failed command: %s\n"), jcr->errmsg);
295 Jmsg(jcr, M_FATAL, 0, _("Bad Use Device command: %s\n"), jcr->errmsg);
296 bnet_fsend(dir, BAD_use, jcr->errmsg);
299 return false; /* ERROR return */
303 * Destroy the Job Control Record and associated
304 * resources (sockets).
306 void stored_free_jcr(JCR *jcr)
308 if (jcr->file_bsock) {
309 bnet_close(jcr->file_bsock);
310 jcr->file_bsock = NULL;
313 free_pool_memory(jcr->job_name);
315 if (jcr->client_name) {
316 free_memory(jcr->client_name);
317 jcr->client_name = NULL;
319 if (jcr->fileset_name) {
320 free_memory(jcr->fileset_name);
322 if (jcr->fileset_md5) {
323 free_memory(jcr->fileset_md5);
329 if (jcr->RestoreBootstrap) {
330 unlink(jcr->RestoreBootstrap);
331 free_pool_memory(jcr->RestoreBootstrap);
332 jcr->RestoreBootstrap = NULL;
334 if (jcr->next_dev || jcr->prev_dev) {
335 Emsg0(M_FATAL, 0, _("In free_jcr(), but still attached to device!!!!\n"));
337 pthread_cond_destroy(&jcr->job_start_wait);