summaryrefslogtreecommitdiffstats
path: root/plugins/check_ntp_peer.c
diff options
context:
space:
mode:
Diffstat (limited to 'plugins/check_ntp_peer.c')
-rw-r--r--plugins/check_ntp_peer.c628
1 files changed, 216 insertions, 412 deletions
diff --git a/plugins/check_ntp_peer.c b/plugins/check_ntp_peer.c
index 164d5190..1c4702c8 100644
--- a/plugins/check_ntp_peer.c
+++ b/plugins/check_ntp_peer.c
@@ -1,6 +1,6 @@
1/****************************************************************************** 1/******************************************************************************
2* 2*
3* Nagios check_ntp plugin 3* Nagios check_ntp_peer plugin
4* 4*
5* License: GPL 5* License: GPL
6* Copyright (c) 2006 sean finney <seanius@seanius.net> 6* Copyright (c) 2006 sean finney <seanius@seanius.net>
@@ -10,11 +10,16 @@
10* 10*
11* Description: 11* Description:
12* 12*
13* This file contains the check_ntp plugin 13* This file contains the check_ntp_peer plugin
14* 14*
15* This plugin to check ntp servers independant of any commandline 15* This plugin checks an NTP server independent of any commandline
16* programs or external libraries. 16* programs or external libraries.
17* 17*
18* Use this plugin to check the health of an NTP server. It supports
19* checking the offset with the sync peer, the jitter and stratum. This
20* plugin will not check the clock offset between the local host and NTP
21* server; please use check_ntp_time for that purpose.
22*
18* 23*
19* License Information: 24* License Information:
20* 25*
@@ -36,7 +41,7 @@
36 41
37*****************************************************************************/ 42*****************************************************************************/
38 43
39const char *progname = "check_ntp"; 44const char *progname = "check_ntp_peer";
40const char *revision = "$Revision$"; 45const char *revision = "$Revision$";
41const char *copyright = "2007"; 46const char *copyright = "2007";
42const char *email = "nagiosplug-devel@lists.sourceforge.net"; 47const char *email = "nagiosplug-devel@lists.sourceforge.net";
@@ -47,51 +52,28 @@ const char *email = "nagiosplug-devel@lists.sourceforge.net";
47 52
48static char *server_address=NULL; 53static char *server_address=NULL;
49static int verbose=0; 54static int verbose=0;
55static int quiet=0;
50static short do_offset=0; 56static short do_offset=0;
51static char *owarn="60"; 57static char *owarn="60";
52static char *ocrit="120"; 58static char *ocrit="120";
59static short do_stratum=0;
60static char *swarn="-1:16";
61static char *scrit="-1:16";
53static short do_jitter=0; 62static short do_jitter=0;
54static char *jwarn="5000"; 63static char *jwarn="-1:5000";
55static char *jcrit="10000"; 64static char *jcrit="-1:10000";
65static int syncsource_found=0;
56 66
57int process_arguments (int, char **); 67int process_arguments (int, char **);
58thresholds *offset_thresholds = NULL; 68thresholds *offset_thresholds = NULL;
59thresholds *jitter_thresholds = NULL; 69thresholds *jitter_thresholds = NULL;
70thresholds *stratum_thresholds = NULL;
60void print_help (void); 71void print_help (void);
61void print_usage (void); 72void print_usage (void);
62 73
63/* number of times to perform each request to get a good average. */
64#define AVG_NUM 4
65
66/* max size of control message data */ 74/* max size of control message data */
67#define MAX_CM_SIZE 468 75#define MAX_CM_SIZE 468
68 76
69/* this structure holds everything in an ntp request/response as per rfc1305 */
70typedef struct {
71 uint8_t flags; /* byte with leapindicator,vers,mode. see macros */
72 uint8_t stratum; /* clock stratum */
73 int8_t poll; /* polling interval */
74 int8_t precision; /* precision of the local clock */
75 int32_t rtdelay; /* total rt delay, as a fixed point num. see macros */
76 uint32_t rtdisp; /* like above, but for max err to primary src */
77 uint32_t refid; /* ref clock identifier */
78 uint64_t refts; /* reference timestamp. local time local clock */
79 uint64_t origts; /* time at which request departed client */
80 uint64_t rxts; /* time at which request arrived at server */
81 uint64_t txts; /* time at which request departed server */
82} ntp_message;
83
84/* this structure holds data about results from querying offset from a peer */
85typedef struct {
86 time_t waiting; /* ts set when we started waiting for a response */
87 int num_responses; /* number of successfully recieved responses */
88 uint8_t stratum; /* copied verbatim from the ntp_message */
89 double rtdelay; /* converted from the ntp_message */
90 double rtdisp; /* converted from the ntp_message */
91 double offset[AVG_NUM]; /* offsets from each response */
92 uint8_t flags; /* byte with leapindicator,vers,mode. see macros */
93} ntp_server_results;
94
95/* this structure holds everything in an ntp control message as per rfc1305 */ 77/* this structure holds everything in an ntp control message as per rfc1305 */
96typedef struct { 78typedef struct {
97 uint8_t flags; /* byte with leapindicator,vers,mode. see macros */ 79 uint8_t flags; /* byte with leapindicator,vers,mode. see macros */
@@ -147,57 +129,6 @@ typedef struct {
147#define PEER_INCLUDED 0x04 129#define PEER_INCLUDED 0x04
148#define PEER_SYNCSOURCE 0x06 130#define PEER_SYNCSOURCE 0x06
149 131
150/**
151 ** a note about the 32-bit "fixed point" numbers:
152 **
153 they are divided into halves, each being a 16-bit int in network byte order:
154 - the first 16 bits are an int on the left side of a decimal point.
155 - the second 16 bits represent a fraction n/(2^16)
156 likewise for the 64-bit "fixed point" numbers with everything doubled :)
157 **/
158
159/* macros to access the left/right 16 bits of a 32-bit ntp "fixed point"
160 number. note that these can be used as lvalues too */
161#define L16(x) (((uint16_t*)&x)[0])
162#define R16(x) (((uint16_t*)&x)[1])
163/* macros to access the left/right 32 bits of a 64-bit ntp "fixed point"
164 number. these too can be used as lvalues */
165#define L32(x) (((uint32_t*)&x)[0])
166#define R32(x) (((uint32_t*)&x)[1])
167
168/* ntp wants seconds since 1/1/00, epoch is 1/1/70. this is the difference */
169#define EPOCHDIFF 0x83aa7e80UL
170
171/* extract a 32-bit ntp fixed point number into a double */
172#define NTP32asDOUBLE(x) (ntohs(L16(x)) + (double)ntohs(R16(x))/65536.0)
173
174/* likewise for a 64-bit ntp fp number */
175#define NTP64asDOUBLE(n) (double)(((uint64_t)n)?\
176 (ntohl(L32(n))-EPOCHDIFF) + \
177 (.00000001*(0.5+(double)(ntohl(R32(n))/42.94967296))):\
178 0)
179
180/* convert a struct timeval to a double */
181#define TVasDOUBLE(x) (double)(x.tv_sec+(0.000001*x.tv_usec))
182
183/* convert an ntp 64-bit fp number to a struct timeval */
184#define NTP64toTV(n,t) \
185 do{ if(!n) t.tv_sec = t.tv_usec = 0; \
186 else { \
187 t.tv_sec=ntohl(L32(n))-EPOCHDIFF; \
188 t.tv_usec=(int)(0.5+(double)(ntohl(R32(n))/4294.967296)); \
189 } \
190 }while(0)
191
192/* convert a struct timeval to an ntp 64-bit fp number */
193#define TVtoNTP64(t,n) \
194 do{ if(!t.tv_usec && !t.tv_sec) n=0x0UL; \
195 else { \
196 L32(n)=htonl(t.tv_sec + EPOCHDIFF); \
197 R32(n)=htonl((uint64_t)((4294.967296*t.tv_usec)+.5)); \
198 } \
199 } while(0)
200
201/* NTP control message header is 12 bytes, plus any data in the data 132/* NTP control message header is 12 bytes, plus any data in the data
202 * field, plus null padding to the nearest 32-bit boundary per rfc. 133 * field, plus null padding to the nearest 32-bit boundary per rfc.
203 */ 134 */
@@ -210,42 +141,6 @@ typedef struct {
210 printf("%u.%u.%u.%u", (x>>24)&0xff, (x>>16)&0xff, (x>>8)&0xff, x&0xff);\ 141 printf("%u.%u.%u.%u", (x>>24)&0xff, (x>>16)&0xff, (x>>8)&0xff, x&0xff);\
211 }while(0); 142 }while(0);
212 143
213/* calculate the offset of the local clock */
214static inline double calc_offset(const ntp_message *m, const struct timeval *t){
215 double client_tx, peer_rx, peer_tx, client_rx;
216 client_tx = NTP64asDOUBLE(m->origts);
217 peer_rx = NTP64asDOUBLE(m->rxts);
218 peer_tx = NTP64asDOUBLE(m->txts);
219 client_rx=TVasDOUBLE((*t));
220 return (.5*((peer_tx-client_rx)+(peer_rx-client_tx)));
221}
222
223/* print out a ntp packet in human readable/debuggable format */
224void print_ntp_message(const ntp_message *p){
225 struct timeval ref, orig, rx, tx;
226
227 NTP64toTV(p->refts,ref);
228 NTP64toTV(p->origts,orig);
229 NTP64toTV(p->rxts,rx);
230 NTP64toTV(p->txts,tx);
231
232 printf("packet contents:\n");
233 printf("\tflags: 0x%.2x\n", p->flags);
234 printf("\t li=%d (0x%.2x)\n", LI(p->flags), p->flags&LI_MASK);
235 printf("\t vn=%d (0x%.2x)\n", VN(p->flags), p->flags&VN_MASK);
236 printf("\t mode=%d (0x%.2x)\n", MODE(p->flags), p->flags&MODE_MASK);
237 printf("\tstratum = %d\n", p->stratum);
238 printf("\tpoll = %g\n", pow(2, p->poll));
239 printf("\tprecision = %g\n", pow(2, p->precision));
240 printf("\trtdelay = %-.16g\n", NTP32asDOUBLE(p->rtdelay));
241 printf("\trtdisp = %-.16g\n", NTP32asDOUBLE(p->rtdisp));
242 printf("\trefid = %x\n", p->refid);
243 printf("\trefts = %-.16g\n", NTP64asDOUBLE(p->refts));
244 printf("\torigts = %-.16g\n", NTP64asDOUBLE(p->origts));
245 printf("\trxts = %-.16g\n", NTP64asDOUBLE(p->rxts));
246 printf("\ttxts = %-.16g\n", NTP64asDOUBLE(p->txts));
247}
248
249void print_ntp_control_message(const ntp_control_message *p){ 144void print_ntp_control_message(const ntp_control_message *p){
250 int i=0, numpeers=0; 145 int i=0, numpeers=0;
251 const ntp_assoc_status_pair *peer=NULL; 146 const ntp_assoc_status_pair *peer=NULL;
@@ -282,222 +177,25 @@ void print_ntp_control_message(const ntp_control_message *p){
282 } 177 }
283} 178}
284 179
285void setup_request(ntp_message *p){ 180char *extract_value(const char *varlist, const char *name){
286 struct timeval t; 181 char *tmpvarlist=NULL, *tmpkey=NULL, *value=NULL;
287 182 int last=0;
288 memset(p, 0, sizeof(ntp_message));
289 LI_SET(p->flags, LI_ALARM);
290 VN_SET(p->flags, 4);
291 MODE_SET(p->flags, MODE_CLIENT);
292 p->poll=4;
293 p->precision=(int8_t)0xfa;
294 L16(p->rtdelay)=htons(1);
295 L16(p->rtdisp)=htons(1);
296
297 gettimeofday(&t, NULL);
298 TVtoNTP64(t,p->txts);
299}
300
301/* select the "best" server from a list of servers, and return its index.
302 * this is done by filtering servers based on stratum, dispersion, and
303 * finally round-trip delay. */
304int best_offset_server(const ntp_server_results *slist, int nservers){
305 int i=0, j=0, cserver=0, candidates[5], csize=0;
306
307 /* for each server */
308 for(cserver=0; cserver<nservers; cserver++){
309 /* sort out servers with error flags */
310 if ( LI(slist[cserver].flags) != LI_NOWARNING ){
311 if (verbose) printf("discarding peer id %d: flags=%d\n", cserver, LI(slist[cserver].flags));
312 break;
313 }
314
315 /* compare it to each of the servers already in the candidate list */
316 for(i=0; i<csize; i++){
317 /* does it have an equal or better stratum? */
318 if(slist[cserver].stratum <= slist[i].stratum){
319 /* does it have an equal or better dispersion? */
320 if(slist[cserver].rtdisp <= slist[i].rtdisp){
321 /* does it have a better rtdelay? */
322 if(slist[cserver].rtdelay < slist[i].rtdelay){
323 break;
324 }
325 }
326 }
327 }
328
329 /* if we haven't reached the current list's end, move everyone
330 * over one to the right, and insert the new candidate */
331 if(i<csize){
332 for(j=5; j>i; j--){
333 candidates[j]=candidates[j-1];
334 }
335 }
336 /* regardless, if they should be on the list... */
337 if(i<5) {
338 candidates[i]=cserver;
339 if(csize<5) csize++;
340 /* otherwise discard the server */
341 } else {
342 DBG(printf("discarding peer id %d\n", cserver));
343 }
344 }
345
346 if(csize>0) {
347 DBG(printf("best server selected: peer %d\n", candidates[0]));
348 return candidates[0];
349 } else {
350 DBG(printf("no peers meeting synchronization criteria :(\n"));
351 return -1;
352 }
353}
354
355/* do everything we need to get the total average offset
356 * - we use a certain amount of parallelization with poll() to ensure
357 * we don't waste time sitting around waiting for single packets.
358 * - we also "manually" handle resolving host names and connecting, because
359 * we have to do it in a way that our lazy macros don't handle currently :( */
360double offset_request(const char *host, int *status){
361 int i=0, j=0, ga_result=0, num_hosts=0, *socklist=NULL, respnum=0;
362 int servers_completed=0, one_written=0, one_read=0, servers_readable=0, best_index=-1;
363 time_t now_time=0, start_ts=0;
364 ntp_message *req=NULL;
365 double avg_offset=0.;
366 struct timeval recv_time;
367 struct addrinfo *ai=NULL, *ai_tmp=NULL, hints;
368 struct pollfd *ufds=NULL;
369 ntp_server_results *servers=NULL;
370
371 /* setup hints to only return results from getaddrinfo that we'd like */
372 memset(&hints, 0, sizeof(struct addrinfo));
373 hints.ai_family = address_family;
374 hints.ai_protocol = IPPROTO_UDP;
375 hints.ai_socktype = SOCK_DGRAM;
376
377 /* fill in ai with the list of hosts resolved by the host name */
378 ga_result = getaddrinfo(host, "123", &hints, &ai);
379 if(ga_result!=0){
380 die(STATE_UNKNOWN, "error getting address for %s: %s\n",
381 host, gai_strerror(ga_result));
382 }
383
384 /* count the number of returned hosts, and allocate stuff accordingly */
385 for(ai_tmp=ai; ai_tmp!=NULL; ai_tmp=ai_tmp->ai_next){ num_hosts++; }
386 req=(ntp_message*)malloc(sizeof(ntp_message)*num_hosts);
387 if(req==NULL) die(STATE_UNKNOWN, "can not allocate ntp message array");
388 socklist=(int*)malloc(sizeof(int)*num_hosts);
389 if(socklist==NULL) die(STATE_UNKNOWN, "can not allocate socket array");
390 ufds=(struct pollfd*)malloc(sizeof(struct pollfd)*num_hosts);
391 if(ufds==NULL) die(STATE_UNKNOWN, "can not allocate socket array");
392 servers=(ntp_server_results*)malloc(sizeof(ntp_server_results)*num_hosts);
393 if(servers==NULL) die(STATE_UNKNOWN, "can not allocate server array");
394 memset(servers, 0, sizeof(ntp_server_results)*num_hosts);
395
396 /* setup each socket for writing, and the corresponding struct pollfd */
397 ai_tmp=ai;
398 for(i=0;ai_tmp;i++){
399 socklist[i]=socket(ai_tmp->ai_family, SOCK_DGRAM, IPPROTO_UDP);
400 if(socklist[i] == -1) {
401 perror(NULL);
402 die(STATE_UNKNOWN, "can not create new socket");
403 }
404 if(connect(socklist[i], ai_tmp->ai_addr, ai_tmp->ai_addrlen)){
405 die(STATE_UNKNOWN, "can't create socket connection");
406 } else {
407 ufds[i].fd=socklist[i];
408 ufds[i].events=POLLIN;
409 ufds[i].revents=0;
410 }
411 ai_tmp = ai_tmp->ai_next;
412 }
413
414 /* now do AVG_NUM checks to each host. we stop before timeout/2 seconds
415 * have passed in order to ensure post-processing and jitter time. */
416 now_time=start_ts=time(NULL);
417 while(servers_completed<num_hosts && now_time-start_ts <= socket_timeout/2){
418 /* loop through each server and find each one which hasn't
419 * been touched in the past second or so and is still lacking
420 * some responses. for each of these servers, send a new request,
421 * and update the "waiting" timestamp with the current time. */
422 one_written=0;
423 now_time=time(NULL);
424
425 for(i=0; i<num_hosts; i++){
426 if(servers[i].waiting<now_time && servers[i].num_responses<AVG_NUM){
427 if(verbose && servers[i].waiting != 0) printf("re-");
428 if(verbose) printf("sending request to peer %d\n", i);
429 setup_request(&req[i]);
430 write(socklist[i], &req[i], sizeof(ntp_message));
431 servers[i].waiting=now_time;
432 one_written=1;
433 break;
434 }
435 }
436
437 /* quickly poll for any sockets with pending data */
438 servers_readable=poll(ufds, num_hosts, 100);
439 if(servers_readable==-1){
440 perror("polling ntp sockets");
441 die(STATE_UNKNOWN, "communication errors");
442 }
443
444 /* read from any sockets with pending data */
445 for(i=0; servers_readable && i<num_hosts; i++){
446 if(ufds[i].revents&POLLIN && servers[i].num_responses < AVG_NUM){
447 if(verbose) {
448 printf("response from peer %d: ", i);
449 }
450 183
451 read(ufds[i].fd, &req[i], sizeof(ntp_message)); 184 /* The following code require a non-empty varlist */
452 gettimeofday(&recv_time, NULL); 185 if(strlen(varlist) == 0)
453 DBG(print_ntp_message(&req[i])); 186 return NULL;
454 respnum=servers[i].num_responses++;
455 servers[i].offset[respnum]=calc_offset(&req[i], &recv_time);
456 if(verbose) {
457 printf("offset %.10g\n", servers[i].offset[respnum]);
458 }
459 servers[i].stratum=req[i].stratum;
460 servers[i].rtdisp=NTP32asDOUBLE(req[i].rtdisp);
461 servers[i].rtdelay=NTP32asDOUBLE(req[i].rtdelay);
462 servers[i].waiting=0;
463 servers[i].flags=req[i].flags;
464 servers_readable--;
465 one_read = 1;
466 if(servers[i].num_responses==AVG_NUM) servers_completed++;
467 }
468 }
469 /* lather, rinse, repeat. */
470 }
471 187
472 if (one_read == 0) { 188 tmpvarlist = strdup(varlist);
473 die(STATE_CRITICAL, "NTP CRITICAL: No response from NTP server\n"); 189 tmpkey = strtok(tmpvarlist, "=");
474 }
475 190
476 /* now, pick the best server from the list */ 191 do {
477 best_index=best_offset_server(servers, num_hosts); 192 if(strstr(tmpkey, name) != NULL) {
478 if(best_index < 0){ 193 value = strtok(NULL, ",");
479 *status=STATE_UNKNOWN; 194 last = 1;
480 } else {
481 /* finally, calculate the average offset */
482 for(i=0; i<servers[best_index].num_responses;i++){
483 avg_offset+=servers[best_index].offset[j];
484 } 195 }
485 avg_offset/=servers[best_index].num_responses; 196 } while (last == 0 && (tmpkey = strtok(NULL, "=")));
486 }
487 197
488 /* cleanup */ 198 return value;
489 /* FIXME: Not closing the socket to avoid re-use of the local port
490 * which can cause old NTP packets to be read instead of NTP control
491 * pactets in jitter_request(). THERE MUST BE ANOTHER WAY...
492 * for(j=0; j<num_hosts; j++){ close(socklist[j]); } */
493 free(socklist);
494 free(ufds);
495 free(servers);
496 free(req);
497 freeaddrinfo(ai);
498
499 if(verbose) printf("overall average offset: %.10g\n", avg_offset);
500 return avg_offset;
501} 199}
502 200
503void 201void
@@ -511,20 +209,36 @@ setup_control_request(ntp_control_message *p, uint8_t opcode, uint16_t seq){
511 /* Remaining fields are zero for requests */ 209 /* Remaining fields are zero for requests */
512} 210}
513 211
514/* XXX handle responses with the error bit set */ 212/* This function does all the actual work; roughly here's what it does
515double jitter_request(const char *host, int *status){ 213 * beside setting the offest, jitter and stratum passed as argument:
516 int conn=-1, i, npeers=0, num_candidates=0, syncsource_found=0; 214 * - offset can be negative, so if it cannot get the offset, offset_result
517 int run=0, min_peer_sel=PEER_INCLUDED, num_selected=0, num_valid=0; 215 * is set to UNKNOWN, otherwise OK.
216 * - jitter and stratum are set to -1 if they cannot be retrieved so any
217 * positive value means a success retrieving the value.
218 * - status is set to WARNING if there's no sync.peer (otherwise OK) and is
219 * the return value of the function.
220 * status is pretty much useless as syncsource_found is a global variable
221 * used later in main to check is the server was synchronized. It works
222 * so I left it alone */
223int ntp_request(const char *host, double *offset, int *offset_result, double *jitter, int *stratum){
224 int conn=-1, i, npeers=0, num_candidates=0;
225 double tmp_offset = 0;
226 int min_peer_sel=PEER_INCLUDED;
518 int peers_size=0, peer_offset=0; 227 int peers_size=0, peer_offset=0;
228 int status;
519 ntp_assoc_status_pair *peers=NULL; 229 ntp_assoc_status_pair *peers=NULL;
520 ntp_control_message req; 230 ntp_control_message req;
521 const char *getvar = "jitter"; 231 const char *getvar = "stratum,offset,jitter";
522 double rval = 0.0, jitter = -1.0; 232 char *data, *value, *nptr;
523 char *startofvalue=NULL, *nptr=NULL;
524 void *tmp; 233 void *tmp;
525 234
235 status = STATE_OK;
236 *offset_result = STATE_UNKNOWN;
237 *jitter = *stratum = -1;
238
526 /* Long-winded explanation: 239 /* Long-winded explanation:
527 * Getting the jitter requires a number of steps: 240 * Getting the sync peer offset, jitter and stratum requires a number of
241 * steps:
528 * 1) Send a READSTAT request. 242 * 1) Send a READSTAT request.
529 * 2) Interpret the READSTAT reply 243 * 2) Interpret the READSTAT reply
530 * a) The data section contains a list of peer identifiers (16 bits) 244 * a) The data section contains a list of peer identifiers (16 bits)
@@ -535,7 +249,8 @@ double jitter_request(const char *host, int *status){
535 * set a minimum of warning. 249 * set a minimum of warning.
536 * 3) Send a READVAR request for information on each peer identified 250 * 3) Send a READVAR request for information on each peer identified
537 * in 2b greater than the minimum selection value. 251 * in 2b greater than the minimum selection value.
538 * 4) Extract the jitter value from the data[] (it's ASCII) 252 * 4) Extract the offset, jitter and stratum value from the data[]
253 * (it's ASCII)
539 */ 254 */
540 my_udp_connect(server_address, 123, &conn); 255 my_udp_connect(server_address, 123, &conn);
541 256
@@ -564,7 +279,7 @@ double jitter_request(const char *host, int *status){
564 } while(req.op&REM_MORE); 279 } while(req.op&REM_MORE);
565 280
566 /* first, let's find out if we have a sync source, or if there are 281 /* first, let's find out if we have a sync source, or if there are
567 * at least some candidates. in the case of the latter we'll issue 282 * at least some candidates. In the latter case we'll issue
568 * a warning but go ahead with the check on them. */ 283 * a warning but go ahead with the check on them. */
569 for (i = 0; i < npeers; i++){ 284 for (i = 0; i < npeers; i++){
570 if (PEER_SEL(peers[i].status) >= PEER_INCLUDED){ 285 if (PEER_SEL(peers[i].status) >= PEER_INCLUDED){
@@ -578,21 +293,22 @@ double jitter_request(const char *host, int *status){
578 if(verbose) printf("%d candiate peers available\n", num_candidates); 293 if(verbose) printf("%d candiate peers available\n", num_candidates);
579 if(verbose && syncsource_found) printf("synchronization source found\n"); 294 if(verbose && syncsource_found) printf("synchronization source found\n");
580 if(! syncsource_found){ 295 if(! syncsource_found){
581 *status = STATE_UNKNOWN; 296 status = STATE_WARNING;
582 if(verbose) printf("warning: no synchronization source found\n"); 297 if(verbose) printf("warning: no synchronization source found\n");
583 } 298 }
584 299
585 300
586 for (run=0; run<AVG_NUM; run++){ 301 for (i = 0; i < npeers; i++){
587 if(verbose) printf("jitter run %d of %d\n", run+1, AVG_NUM); 302 /* Only query this server if it is the current sync source */
588 for (i = 0; i < npeers; i++){ 303 /* If there's no sync.peer, query all candidates and use the best one */
589 /* Only query this server if it is the current sync source */ 304 if (PEER_SEL(peers[i].status) >= min_peer_sel){
590 if (PEER_SEL(peers[i].status) >= min_peer_sel){ 305 if(verbose) printf("Getting offset, jitter and stratum for peer %.2x\n", ntohs(peers[i].assoc));
591 num_selected++; 306 asprintf(&data, "");
307 do{
592 setup_control_request(&req, OP_READVAR, 2); 308 setup_control_request(&req, OP_READVAR, 2);
593 req.assoc = peers[i].assoc; 309 req.assoc = peers[i].assoc;
594 /* By spec, putting the variable name "jitter" in the request 310 /* Putting the wanted variable names in the request
595 * should cause the server to provide _only_ the jitter value. 311 * cause the server to provide _only_ the requested values.
596 * thus reducing net traffic, guaranteeing us only a single 312 * thus reducing net traffic, guaranteeing us only a single
597 * datagram in reply, and making intepretation much simpler 313 * datagram in reply, and making intepretation much simpler
598 */ 314 */
@@ -605,48 +321,97 @@ double jitter_request(const char *host, int *status){
605 DBG(print_ntp_control_message(&req)); 321 DBG(print_ntp_control_message(&req));
606 322
607 req.count = htons(MAX_CM_SIZE); 323 req.count = htons(MAX_CM_SIZE);
608 DBG(printf("recieving READVAR response...\n")); 324 DBG(printf("receiving READVAR response...\n"));
609 read(conn, &req, SIZEOF_NTPCM(req)); 325 read(conn, &req, SIZEOF_NTPCM(req));
610 DBG(print_ntp_control_message(&req)); 326 DBG(print_ntp_control_message(&req));
611 327
612 if(req.op&REM_ERROR && strstr(getvar, "jitter")) { 328 if(!(req.op&REM_ERROR))
613 if(verbose) printf("The 'jitter' command failed (old ntp server?)\nRestarting with 'dispersion'...\n"); 329 asprintf(&data, "%s%s", data, req.data);
614 getvar = "dispersion"; 330 } while(req.op&REM_MORE);
615 num_selected--; 331
332 if(req.op&REM_ERROR) {
333 if(strstr(getvar, "jitter")) {
334 if(verbose) printf("The command failed. This is usually caused by servers refusing the 'jitter'\nvariable. Restarting with 'dispersion'...\n");
335 getvar = "stratum,offset,dispersion";
336 i--;
337 continue;
338 } else if(strlen(getvar)) {
339 if(verbose) printf("Server didn't like dispersion either; will retrieve everything\n");
340 getvar = "";
616 i--; 341 i--;
617 continue; 342 continue;
618 } 343 }
344 }
345
346 if(verbose > 1)
347 printf("Server responded: >>>%s<<<\n", data);
348
349 /* get the offset */
350 if(verbose)
351 printf("parsing offset from peer %.2x: ", ntohs(peers[i].assoc));
352
353 value = extract_value(data, "offset");
354 nptr=NULL;
355 /* Convert the value if we have one */
356 if(value != NULL)
357 tmp_offset = strtod(value, &nptr) / 1000;
358 /* If value is null or no conversion was performed */
359 if(value == NULL || value==nptr) {
360 if(verbose) printf("error: unable to read server offset response.\n");
361 } else {
362 if(verbose) printf("%.10g\n", tmp_offset);
363 if(*offset_result == STATE_UNKNOWN || fabs(tmp_offset) < fabs(*offset)) {
364 *offset = tmp_offset;
365 *offset_result = STATE_OK;
366 } else {
367 /* Skip this one; move to the next */
368 continue;
369 }
370 }
619 371
620 /* get to the float value */ 372 if(do_jitter) {
373 /* get the jitter */
621 if(verbose) { 374 if(verbose) {
622 printf("parsing jitter from peer %.2x: ", ntohs(peers[i].assoc)); 375 printf("parsing %s from peer %.2x: ", strstr(getvar, "dispersion") != NULL ? "dispersion" : "jitter", ntohs(peers[i].assoc));
623 } 376 }
624 startofvalue = strchr(req.data, '='); 377 value = extract_value(data, strstr(getvar, "dispersion") != NULL ? "dispersion" : "jitter");
625 if(startofvalue != NULL) { 378 nptr=NULL;
626 startofvalue++; 379 /* Convert the value if we have one */
627 jitter = strtod(startofvalue, &nptr); 380 if(value != NULL)
381 *jitter = strtod(value, &nptr);
382 /* If value is null or no conversion was performed */
383 if(value == NULL || value==nptr) {
384 if(verbose) printf("error: unable to read server jitter/dispersion response.\n");
385 *jitter = -1;
386 } else if(verbose) {
387 printf("%.10g\n", *jitter);
388 }
389 }
390
391 if(do_stratum) {
392 /* get the stratum */
393 if(verbose) {
394 printf("parsing stratum from peer %.2x: ", ntohs(peers[i].assoc));
628 } 395 }
629 if(startofvalue == NULL || startofvalue==nptr){ 396 value = extract_value(data, "stratum");
630 printf("warning: unable to read server jitter response.\n"); 397 nptr=NULL;
631 *status = STATE_UNKNOWN; 398 /* Convert the value if we have one */
399 if(value != NULL)
400 *stratum = strtol(value, &nptr, 10);
401 if(value == NULL || value==nptr) {
402 if(verbose) printf("error: unable to read server stratum response.\n");
403 *stratum = -1;
632 } else { 404 } else {
633 if(verbose) printf("%g\n", jitter); 405 if(verbose) printf("%i\n", *stratum);
634 num_valid++;
635 rval += jitter;
636 } 406 }
637 } 407 }
638 } 408 } /* if (PEER_SEL(peers[i].status) >= min_peer_sel) */
639 if(verbose){ 409 } /* for (i = 0; i < npeers; i++) */
640 printf("jitter parsed from %d/%d peers\n", num_valid, num_selected);
641 }
642 }
643
644 rval = num_valid ? rval / num_valid : -1.0;
645 410
646 close(conn); 411 close(conn);
647 if(peers!=NULL) free(peers); 412 if(peers!=NULL) free(peers);
648 /* If we return -1.0, it means no synchronization source was found */ 413
649 return rval; 414 return status;
650} 415}
651 416
652int process_arguments(int argc, char **argv){ 417int process_arguments(int argc, char **argv){
@@ -658,8 +423,11 @@ int process_arguments(int argc, char **argv){
658 {"verbose", no_argument, 0, 'v'}, 423 {"verbose", no_argument, 0, 'v'},
659 {"use-ipv4", no_argument, 0, '4'}, 424 {"use-ipv4", no_argument, 0, '4'},
660 {"use-ipv6", no_argument, 0, '6'}, 425 {"use-ipv6", no_argument, 0, '6'},
426 {"quiet", no_argument, 0, 'q'},
661 {"warning", required_argument, 0, 'w'}, 427 {"warning", required_argument, 0, 'w'},
662 {"critical", required_argument, 0, 'c'}, 428 {"critical", required_argument, 0, 'c'},
429 {"swarn", required_argument, 0, 'W'},
430 {"scrit", required_argument, 0, 'C'},
663 {"jwarn", required_argument, 0, 'j'}, 431 {"jwarn", required_argument, 0, 'j'},
664 {"jcrit", required_argument, 0, 'k'}, 432 {"jcrit", required_argument, 0, 'k'},
665 {"timeout", required_argument, 0, 't'}, 433 {"timeout", required_argument, 0, 't'},
@@ -672,7 +440,7 @@ int process_arguments(int argc, char **argv){
672 usage ("\n"); 440 usage ("\n");
673 441
674 while (1) { 442 while (1) {
675 c = getopt_long (argc, argv, "Vhv46w:c:j:k:t:H:", longopts, &option); 443 c = getopt_long (argc, argv, "Vhv46qw:c:W:C:j:k:t:H:", longopts, &option);
676 if (c == -1 || c == EOF || c == 1) 444 if (c == -1 || c == EOF || c == 1)
677 break; 445 break;
678 446
@@ -688,6 +456,9 @@ int process_arguments(int argc, char **argv){
688 case 'v': 456 case 'v':
689 verbose++; 457 verbose++;
690 break; 458 break;
459 case 'q':
460 quiet = 1;
461 break;
691 case 'w': 462 case 'w':
692 do_offset=1; 463 do_offset=1;
693 owarn = optarg; 464 owarn = optarg;
@@ -696,6 +467,14 @@ int process_arguments(int argc, char **argv){
696 do_offset=1; 467 do_offset=1;
697 ocrit = optarg; 468 ocrit = optarg;
698 break; 469 break;
470 case 'W':
471 do_stratum=1;
472 swarn = optarg;
473 break;
474 case 'C':
475 do_stratum=1;
476 scrit = optarg;
477 break;
699 case 'j': 478 case 'j':
700 do_jitter=1; 479 do_jitter=1;
701 jwarn = optarg; 480 jwarn = optarg;
@@ -746,24 +525,31 @@ char *perfd_offset (double offset)
746 525
747char *perfd_jitter (double jitter) 526char *perfd_jitter (double jitter)
748{ 527{
749 return fperfdata ("jitter", jitter, "s", 528 return fperfdata ("jitter", jitter, "",
750 do_jitter, jitter_thresholds->warning->end, 529 do_jitter, jitter_thresholds->warning->end,
751 do_jitter, jitter_thresholds->critical->end, 530 do_jitter, jitter_thresholds->critical->end,
752 TRUE, 0, FALSE, 0); 531 TRUE, 0, FALSE, 0);
753} 532}
754 533
534char *perfd_stratum (int stratum)
535{
536 return perfdata ("stratum", stratum, "",
537 do_stratum, (int)stratum_thresholds->warning->end,
538 do_stratum, (int)stratum_thresholds->critical->end,
539 TRUE, 0, TRUE, 16);
540}
541
755int main(int argc, char *argv[]){ 542int main(int argc, char *argv[]){
756 int result, offset_result, jitter_result; 543 int result, offset_result, stratum;
757 double offset=0, jitter=0; 544 double offset=0, jitter=0;
758 char *result_line, *perfdata_line; 545 char *result_line, *perfdata_line;
759 546
760 result = offset_result = jitter_result = STATE_OK;
761
762 if (process_arguments (argc, argv) == ERROR) 547 if (process_arguments (argc, argv) == ERROR)
763 usage4 (_("Could not parse arguments")); 548 usage4 (_("Could not parse arguments"));
764 549
765 set_thresholds(&offset_thresholds, owarn, ocrit); 550 set_thresholds(&offset_thresholds, owarn, ocrit);
766 set_thresholds(&jitter_thresholds, jwarn, jcrit); 551 set_thresholds(&jitter_thresholds, jwarn, jcrit);
552 set_thresholds(&stratum_thresholds, swarn, scrit);
767 553
768 /* initialize alarm signal handling */ 554 /* initialize alarm signal handling */
769 signal (SIGALRM, socket_timeout_alarm_handler); 555 signal (SIGALRM, socket_timeout_alarm_handler);
@@ -771,29 +557,24 @@ int main(int argc, char *argv[]){
771 /* set socket timeout */ 557 /* set socket timeout */
772 alarm (socket_timeout); 558 alarm (socket_timeout);
773 559
774 offset = offset_request(server_address, &offset_result); 560 /* This returns either OK or WARNING (See comment preceeding ntp_request) */
775 /* check_ntp used to always return CRITICAL if offset_result == STATE_UNKNOWN. 561 result = ntp_request(server_address, &offset, &offset_result, &jitter, &stratum);
776 * Now we'll only do that is the offset thresholds were set */ 562
777 if (do_offset && offset_result == STATE_UNKNOWN) { 563 if(offset_result == STATE_UNKNOWN) {
778 result = STATE_CRITICAL; 564 /* if there's no sync peer (this overrides ntp_request output): */
565 result = (quiet == 1 ? STATE_UNKNOWN : STATE_CRITICAL);
779 } else { 566 } else {
780 result = get_status(fabs(offset), offset_thresholds); 567 /* Be quiet if there's no candidates either */
568 if (quiet == 1 && result == STATE_WARNING)
569 result = STATE_UNKNOWN;
570 result = max_state_alt(result, get_status(fabs(offset), offset_thresholds));
781 } 571 }
782 572
783 /* If not told to check the jitter, we don't even send packets. 573 if(do_stratum)
784 * jitter is checked using NTP control packets, which not all 574 result = max_state_alt(result, get_status(stratum, stratum_thresholds));
785 * servers recognize. Trying to check the jitter on OpenNTPD 575
786 * (for example) will result in an error 576 if(do_jitter)
787 */
788 if(do_jitter){
789 jitter=jitter_request(server_address, &jitter_result);
790 result = max_state_alt(result, get_status(jitter, jitter_thresholds)); 577 result = max_state_alt(result, get_status(jitter, jitter_thresholds));
791 /* -1 indicates that we couldn't calculate the jitter
792 * Only overrides STATE_OK from the offset */
793 if(jitter == -1.0 && result == STATE_OK)
794 result = STATE_UNKNOWN;
795 }
796 result = max_state_alt(result, jitter_result);
797 578
798 switch (result) { 579 switch (result) {
799 case STATE_CRITICAL : 580 case STATE_CRITICAL :
@@ -809,6 +590,9 @@ int main(int argc, char *argv[]){
809 asprintf(&result_line, "NTP UNKNOWN:"); 590 asprintf(&result_line, "NTP UNKNOWN:");
810 break; 591 break;
811 } 592 }
593 if(!syncsource_found)
594 asprintf(&result_line, "%s %s,", result_line, _("Server not synchronized"));
595
812 if(offset_result == STATE_UNKNOWN){ 596 if(offset_result == STATE_UNKNOWN){
813 asprintf(&result_line, "%s %s", result_line, _("Offset unknown")); 597 asprintf(&result_line, "%s %s", result_line, _("Offset unknown"));
814 asprintf(&perfdata_line, ""); 598 asprintf(&perfdata_line, "");
@@ -818,7 +602,11 @@ int main(int argc, char *argv[]){
818 } 602 }
819 if (do_jitter) { 603 if (do_jitter) {
820 asprintf(&result_line, "%s, jitter=%f", result_line, jitter); 604 asprintf(&result_line, "%s, jitter=%f", result_line, jitter);
821 asprintf(&perfdata_line, "%s %s", perfdata_line, perfd_jitter(jitter)); 605 asprintf(&perfdata_line, "%s %s", perfdata_line, perfd_jitter(jitter));
606 }
607 if (do_stratum) {
608 asprintf(&result_line, "%s, stratum=%i", result_line, stratum);
609 asprintf(&perfdata_line, "%s %s", perfdata_line, perfd_stratum(stratum));
822 } 610 }
823 printf("%s|%s\n", result_line, perfdata_line); 611 printf("%s|%s\n", result_line, perfdata_line);
824 612
@@ -833,18 +621,24 @@ void print_help(void){
833 621
834 printf ("Copyright (c) 2006 Sean Finney\n"); 622 printf ("Copyright (c) 2006 Sean Finney\n");
835 printf (COPYRIGHT, copyright, email); 623 printf (COPYRIGHT, copyright, email);
836
837 printf ("%s\n", _("This plugin checks the selected ntp server"));
838 624
839 printf ("\n\n"); 625 printf ("%s\n", _("This plugin checks the selected ntp server"));
840 626
627 printf ("\n\n");
628
841 print_usage(); 629 print_usage();
842 printf (_(UT_HELP_VRSN)); 630 printf (_(UT_HELP_VRSN));
843 printf (_(UT_HOST_PORT), 'p', "123"); 631 printf (_(UT_HOST_PORT), 'p', "123");
632 printf (" %s\n", "-q, --quiet");
633 printf (" %s\n", _("Returns UNKNOWN instead of CRITICAL or WARNING if server isn't synchronized"));
844 printf (" %s\n", "-w, --warning=THRESHOLD"); 634 printf (" %s\n", "-w, --warning=THRESHOLD");
845 printf (" %s\n", _("Offset to result in warning status (seconds)")); 635 printf (" %s\n", _("Offset to result in warning status (seconds)"));
846 printf (" %s\n", "-c, --critical=THRESHOLD"); 636 printf (" %s\n", "-c, --critical=THRESHOLD");
847 printf (" %s\n", _("Offset to result in critical status (seconds)")); 637 printf (" %s\n", _("Offset to result in critical status (seconds)"));
638 printf (" %s\n", "-W, --warning=THRESHOLD");
639 printf (" %s\n", _("Warning threshold for stratum"));
640 printf (" %s\n", "-W, --critical=THRESHOLD");
641 printf (" %s\n", _("Critical threshold for stratum"));
848 printf (" %s\n", "-j, --warning=THRESHOLD"); 642 printf (" %s\n", "-j, --warning=THRESHOLD");
849 printf (" %s\n", _("Warning threshold for jitter")); 643 printf (" %s\n", _("Warning threshold for jitter"));
850 printf (" %s\n", "-k, --critical=THRESHOLD"); 644 printf (" %s\n", "-k, --critical=THRESHOLD");
@@ -854,17 +648,26 @@ void print_help(void){
854 648
855 printf("\n"); 649 printf("\n");
856 printf("%s\n", _("Notes:")); 650 printf("%s\n", _("Notes:"));
651 printf(" %s\n", _("This plugin checks an NTP server independent of any commandline"));
652 printf(" %s\n\n", _("programs or external libraries."));
653 printf(" %s\n", _("Use this plugin to check the health of an NTP server. It supports"));
654 printf(" %s\n", _("checking the offset with the sync peer, the jitter and stratum. This"));
655 printf(" %s\n", _("plugin will not check the clock offset between the local host and NTP"));
656 printf(" %s\n\n", _("server; please use check_ntp_time for that purpose."));
657
857 printf(" %s\n", _("See:")); 658 printf(" %s\n", _("See:"));
858 printf(" %s\n", ("http://nagiosplug.sourceforge.net/developer-guidelines.html#THRESHOLDFORMAT")); 659 printf(" %s\n", ("http://nagiosplug.sourceforge.net/developer-guidelines.html#THRESHOLDFORMAT"));
859 printf(" %s\n", _("for THRESHOLD format and examples.")); 660 printf(" %s\n", _("for THRESHOLD format and examples."));
860 661
861 printf("\n"); 662 printf("\n");
862 printf("%s\n", _("Examples:")); 663 printf("%s\n", _("Examples:"));
863 printf(" %s\n", _("Normal offset check:")); 664 printf(" %s\n", _("Normal NTP server check:"));
864 printf(" %s\n", ("./check_ntp -H ntpserv -w 0.5 -c 1")); 665 printf(" %s\n", ("./check_ntp_peer -H ntpserv -w 0.5 -c 1"));
865 printf(" %s\n", _("Check jitter too, avoiding critical notifications if jitter isn't available")); 666 printf(" %s\n", _("Check jitter too, avoiding critical notifications if jitter isn't available"));
866 printf(" %s\n", _("(See Notes above for more details on thresholds formats):")); 667 printf(" %s\n", _("(See Notes above for more details on thresholds formats):"));
867 printf(" %s\n", ("./check_ntp -H ntpserv -w 0.5 -c 1 -j -1:100 -k -1:200")); 668 printf(" %s\n", ("./check_ntp_peer -H ntpserv -w 0.5 -c 1 -j -1:100 -k -1:200"));
669 printf(" %s\n", _("Check only stratum:"));
670 printf(" %s\n", ("./check_ntp_peer -H ntpserv -W 4 -C 6"));
868 671
869 printf (_(UT_SUPPORT)); 672 printf (_(UT_SUPPORT));
870} 673}
@@ -872,6 +675,7 @@ void print_help(void){
872void 675void
873print_usage(void) 676print_usage(void)
874{ 677{
875 printf (_("Usage:")); 678 printf (_("Usage:"));
876 printf(" %s -H <host> [-w <warn>] [-c <crit>] [-j <warn>] [-k <crit>] [-v verbose]\n", progname); 679 printf(" %s -H <host> [-w <warn>] [-c <crit>] [-W <warn>] [-C <crit>]\n", progname);
680 printf(" [-j <warn>] [-k <crit>] [-v verbose]\n");
877} 681}