-rw-r--r-- | src/process.cc | 118 |
1 files changed, 107 insertions, 11 deletions
diff --git a/src/process.cc b/src/process.cc index 1ffac9f..8a5b5d2 100644 --- a/src/process.cc +++ b/src/process.cc | |||
@@ -1,35 +1,51 @@ | |||
1 | #include <stdio.h> | 1 | #include <stdio.h> |
2 | #include <sys/types.h> | 2 | #include <sys/types.h> |
3 | #include <unistd.h> | 3 | #include <unistd.h> |
4 | #include <signal.h> | 4 | #include <signal.h> |
5 | #include <pwd.h> | 5 | #include <pwd.h> |
6 | #include <grp.h> | 6 | #include <grp.h> |
7 | #include <dirent.h> | ||
7 | #include <sys/wait.h> | 8 | #include <sys/wait.h> |
8 | #include <syslog.h> | 9 | #include <syslog.h> |
9 | #include <errno.h> | 10 | #include <errno.h> |
10 | #include <iostream> | 11 | #include <iostream> |
11 | #include <fstream> | 12 | #include <fstream> |
13 | #include <sstream> | ||
12 | #include <stdexcept> | 14 | #include <stdexcept> |
15 | #include <string> | ||
16 | #include <vector> | ||
13 | using namespace std; | 17 | using namespace std; |
14 | #include "process.h" | 18 | #include "process.h" |
15 | #include "configuration.h" | 19 | #include "configuration.h" |
16 | 20 | ||
21 | static multimap<string,pid_t> procpids; | ||
22 | |||
23 | void process::check() const { | ||
24 | if(!pidfile.empty()) { | ||
25 | signal(0); | ||
26 | }else if(!process_name.empty()) { | ||
27 | if(procpids.empty()) | ||
28 | gather_proc_info(); | ||
29 | if(procpids.find(process_name)==procpids.end()) | ||
30 | throw runtime_error("no such process"); | ||
31 | } // XXX: or else? | ||
32 | } | ||
17 | void process::check(const string& id,configuration& config) { | 33 | void process::check(const string& id,configuration& config) { |
18 | try { | 34 | try { |
19 | signal(0); | 35 | check(); |
20 | patience = 0; | 36 | patience = 0; |
21 | }catch(exception& e) { | 37 | }catch(exception& e) { |
22 | if(patience>60) { // TODO: configurable | 38 | if(patience>60) { // TODO: configurable |
23 | patience = 0; | 39 | patience = 0; |
24 | }else{ | 40 | }else{ |
25 | if(patience<10) { // TODO: configurable | 41 | if(patience<10) { // TODO: configurable |
26 | syslog(LOG_NOTICE,"The process '%s' is down, trying to launch.",id.c_str()); | 42 | syslog(LOG_NOTICE,"The process '%s' is down, trying to launch.",id.c_str()); |
27 | do_notify(id,"Starting up", | 43 | do_notify(id,"Starting up", |
28 | "The named process seems to be down. Dudki will try\n" | 44 | "The named process seems to be down. Dudki will try\n" |
29 | "to revive it by running the specified command.\n", | 45 | "to revive it by running the specified command.\n", |
30 | config); | 46 | config); |
31 | try { | 47 | try { |
32 | launch(id,config); | 48 | launch(id,config); |
33 | }catch(exception& e) { | 49 | }catch(exception& e) { |
34 | syslog(LOG_ERR,"Error trying to launch process '%s': %s",id.c_str(),e.what()); | 50 | syslog(LOG_ERR,"Error trying to launch process '%s': %s",id.c_str(),e.what()); |
35 | } | 51 | } |
@@ -163,27 +179,107 @@ void process::notify_mailto(const string& email,const string& id,const string& e | |||
163 | fprintf(mta,"%s: %s\n",i->first.c_str(),i->second.c_str()); | 179 | fprintf(mta,"%s: %s\n",i->first.c_str(),i->second.c_str()); |
164 | } | 180 | } |
165 | fprintf(mta, | 181 | fprintf(mta, |
166 | "Subject: [%s] %s\n\n" | 182 | "Subject: [%s] %s\n\n" |
167 | "%s\n" | 183 | "%s\n" |
168 | "---\n" | 184 | "---\n" |
169 | "This message was sent automatically by the 'dudki' daemon\n", | 185 | "This message was sent automatically by the 'dudki' daemon\n", |
170 | id.c_str(), event.c_str(), | 186 | id.c_str(), event.c_str(), |
171 | description.c_str() ); | 187 | description.c_str() ); |
172 | fclose(mta); | 188 | fclose(mta); |
173 | int status; | 189 | int status; |
174 | waitpid(pid,&status,0); | 190 | waitpid(pid,&status,0); |
175 | // TODO: check the return code | 191 | // TODO: check the return code |
176 | } | 192 | } |
177 | 193 | ||
178 | void process::signal(int signum) const { | 194 | void process::signal(int signum) const { |
179 | ifstream pids(pidfile.c_str(),ios::in); | 195 | if(!pidfile.empty()) { |
180 | if(!pids) | 196 | ifstream pids(pidfile.c_str(),ios::in); |
181 | throw runtime_error("no pidfile found"); | 197 | if(!pids) |
182 | pid_t pid = 0; | 198 | throw runtime_error("no pidfile found"); |
183 | pids >> pid; | 199 | pid_t pid = 0; |
184 | pids.close(); | 200 | pids >> pid; |
185 | if(!pid) | 201 | pids.close(); |
186 | throw runtime_error("no pid in pidfile"); | 202 | if(!pid) |
187 | if(kill(pid,signum)) | 203 | throw runtime_error("no pid in pidfile"); |
188 | throw runtime_error("failed to signal process"); | 204 | if(kill(pid,signum)) |
205 | throw runtime_error("failed to signal process"); | ||
206 | }else if(!process_name.empty()) { | ||
207 | if(procpids.empty()) | ||
208 | gather_proc_info(); | ||
209 | pair<multimap<string,pid_t>::const_iterator,multimap<string,pid_t>::const_iterator> range = procpids.equal_range(process_name); | ||
210 | int count = 0; | ||
211 | for(multimap<string,pid_t>::const_iterator i=range.first;i!=range.second;++i) { | ||
212 | pid_t pid = i->second; | ||
213 | if(kill(i->second,signum)) | ||
214 | throw runtime_error("failed to signal process"); | ||
215 | count++; | ||
216 | } | ||
217 | if(!count) | ||
218 | throw runtime_error("no running instance detected"); | ||
219 | }else | ||
220 | throw runtime_error("nothing is known about the process"); | ||
221 | } | ||
222 | |||
223 | void process::prepare_herd() { | ||
224 | procpids.clear(); | ||
225 | } | ||
226 | void process::unprepare_herd() { | ||
227 | procpids.clear(); | ||
228 | } | ||
229 | void process::gather_proc_info() { | ||
230 | vector<pid_t> allpids; | ||
231 | DIR *pd = opendir("/proc"); | ||
232 | if(!pd) | ||
233 | throw runtime_error("failed to open /proc"); | ||
234 | struct dirent *pde; | ||
235 | pid_t selfpid = getpid(); | ||
236 | while(pde=readdir(pd)) { | ||
237 | errno=0; | ||
238 | pid_t pid = atoi(pde->d_name); | ||
239 | if((!pid) || pid==selfpid) | ||
240 | continue; | ||
241 | allpids.push_back(pid); | ||
242 | } | ||
243 | closedir(pd); | ||
244 | char s[256]; | ||
245 | procpids.clear(); | ||
246 | for(vector<pid_t>::const_iterator i=allpids.begin();i!=allpids.end();++i) { | ||
247 | int r = snprintf(s,sizeof(s),"/proc/%d/stat",*i); | ||
248 | if(r>=sizeof(s) || r<1) | ||
249 | continue; | ||
250 | string cmd; | ||
251 | ifstream ss(s,ios::in); | ||
252 | if(!ss) | ||
253 | continue; | ||
254 | getline(ss,cmd); | ||
255 | string::size_type op = cmd.find('('); | ||
256 | if(op==string::npos) | ||
257 | continue; | ||
258 | cmd.erase(0,op+1); | ||
259 | string::size_type cp = cmd.find(')'); | ||
260 | if(cp==string::npos) | ||
261 | continue; | ||
262 | cmd.erase(cp); | ||
263 | r = snprintf(s,sizeof(s),"/proc/%d/cmdline",*i); | ||
264 | if(r>=sizeof(s) || r<1) | ||
265 | continue; | ||
266 | ifstream cs(s,ios::binary); | ||
267 | if(!cs) | ||
268 | continue; | ||
269 | string command; | ||
270 | while(cs) { | ||
271 | string cl; | ||
272 | getline(cs,cl,(char)0); | ||
273 | string::size_type lsl = cl.rfind('/'); | ||
274 | if(lsl!=string::npos) | ||
275 | cl.erase(0,lsl+1); | ||
276 | if(cl.substr(0,cmd.length())==cmd) { | ||
277 | command = cl; | ||
278 | break; | ||
279 | } | ||
280 | } | ||
281 | procpids.insert(pair<string,pid_t>(cmd,*i)); | ||
282 | if((!command.empty()) && cmd!=command) | ||
283 | procpids.insert(pair<string,pid_t>(command,*i)); | ||
284 | } | ||
189 | } | 285 | } |