Rooba
/
boinc
mirror of https://github.com/BOINC/boinc.git
1
0
Fork 0
Code Issues 1 Packages Projects Releases Wiki Activity
boinc/sched/transitioner.cpp

716 lines
26 KiB
C++
Raw Blame History

// This file is part of BOINC.
// http://boinc.berkeley.edu
// Copyright (C) 2008 University of California
//
// BOINC is free software; you can redistribute it and/or modify it
// under the terms of the GNU Lesser General Public License
// as published by the Free Software Foundation,
// either version 3 of the License, or (at your option) any later version.
//
// BOINC is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
// See the GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with BOINC. If not, see <http://www.gnu.org/licenses/>.
// transitioner - handle transitions in the state of a WU
// - a result has become DONE (via timeout or client reply)
// - the WU error mask is set (e.g. by validater)
// - assimilation is finished
//
// cmdline:
// [ -one_pass ] do one pass, then exit
// [ -d x ] debug level x
// [ -mod n i ] process only WUs with (id mod n) == i
using namespace std;
#include "config.h"
#include <vector>
#include <unistd.h>
#include <cstring>
#include <climits>
#include <cstdlib>
#include <string>
#include <signal.h>
#include <sys/time.h>
#include "boinc_db.h"
#include "util.h"
#include "backend_lib.h"
#include "common_defs.h"
#include "error_numbers.h"
#include "sched_config.h"
#include "sched_util.h"
#include "sched_msgs.h"
#ifdef GCL_SIMULATOR
#include "gcl_simulator.h"
#endif
#define LOCKFILE "transitioner.out"
#define PIDFILE "transitioner.pid"
#define SELECT_LIMIT 1000
#ifdef EINSTEIN_AT_HOME
#define SLEEP_INTERVAL 1
#else
#define SLEEP_INTERVAL 5
#endif
int startup_time;
R_RSA_PRIVATE_KEY key;
int mod_n, mod_i;
bool do_mod = false;
bool one_pass = false;
void signal_handler(int) {
log_messages.printf(MSG_NORMAL, "Signaled by simulator\n");
return;
}
int result_suffix(char* name) {
char* p = strrchr(name, '_');
if (p) return atoi(p+1);
return 0;
}
// A result just timed out.
// Update the host's avg_turnaround and max_results_day.
//
int penalize_host(int hostid, double delay_bound) {
DB_HOST host;
char buf[256];
int retval = host.lookup_id(hostid);
if (retval) return retval;
compute_avg_turnaround(host, delay_bound);
if (host.max_results_day == 0 || host.max_results_day > config.daily_result_quota) {
host.max_results_day = config.daily_result_quota;
}
host.max_results_day -= 1;
if (host.max_results_day < 1) {
host.max_results_day = 1;
}
sprintf(buf,
"avg_turnaround=%f, max_results_day=%d",
host.avg_turnaround, host.max_results_day
);
return host.update_field(buf);
}
int handle_wu(
DB_TRANSITIONER_ITEM_SET& transitioner,
std::vector<TRANSITIONER_ITEM>& items
) {
int ntotal, nerrors, retval, ninprogress, nsuccess;
int nunsent, ncouldnt_send, nover, ndidnt_need, nno_reply;
int canonical_result_index, j;
char suffix[256];
time_t now = time(0), x;
bool all_over_and_validated, have_new_result_to_validate, do_delete;
unsigned int i;
TRANSITIONER_ITEM& wu_item = items[0];
TRANSITIONER_ITEM wu_item_original = wu_item;
// "assigned" WUs aren't supposed to pass through the transitioner.
// If we get one, it's an error
//
if (config.enable_assignment && strstr(wu_item.name, ASSIGNED_WU_STR)) {
DB_WORKUNIT wu;
char buf[256];
wu.id = wu_item.id;
log_messages.printf(MSG_CRITICAL,
"Assigned WU %d unexpectedly found by transitioner\n", wu.id
);
sprintf(buf, "transition_time=%d", INT_MAX);
retval = wu.update_field(buf);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"update_field failed %d\n", retval
);
}
return 0;
}
// count up the number of results in various states,
// and check for timed-out results
//
ntotal = 0;
nunsent = 0;
ninprogress = 0;
nover = 0;
nerrors = 0;
nsuccess = 0;
ncouldnt_send = 0;
nno_reply = 0;
ndidnt_need = 0;
have_new_result_to_validate = false;
int rs, max_result_suffix = -1;
// Scan the WU's results, and find the canonical result if there is one
//
canonical_result_index = -1;
if (wu_item.canonical_resultid) {
for (i=0; i<items.size(); i++) {
TRANSITIONER_ITEM& res_item = items[i];
if (!res_item.res_id) continue;
if (res_item.res_id == wu_item.canonical_resultid) {
canonical_result_index = i;
}
}
}
if (wu_item.canonical_resultid && (canonical_result_index == -1)) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] can't find canonical result\n",
wu_item.id, wu_item.name
);
}
// if there is a canonical result, see if its file are deleted
//
bool canonical_result_files_deleted = false;
if (canonical_result_index >= 0) {
TRANSITIONER_ITEM& cr = items[canonical_result_index];
if (cr.res_file_delete_state == FILE_DELETE_DONE) {
canonical_result_files_deleted = true;
}
}
// Scan this WU's results, and
// 1) count those in various server states;
// 2) identify time-out results and update their server state and outcome
// 3) find the max result suffix (in case need to generate new ones)
// 4) see if we have a new result to validate
// (outcome SUCCESS and validate_state INIT)
//
for (i=0; i<items.size(); i++) {
TRANSITIONER_ITEM& res_item = items[i];
if (!res_item.res_id) continue;
ntotal++;
rs = result_suffix(res_item.res_name);
if (rs > max_result_suffix) max_result_suffix = rs;
switch (res_item.res_server_state) {
case RESULT_SERVER_STATE_UNSENT:
nunsent++;
break;
case RESULT_SERVER_STATE_IN_PROGRESS:
if (res_item.res_report_deadline < now) {
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] [RESULT#%d %s] result timed out (%d < %d) server_state:IN_PROGRESS=>OVER; outcome:NO_REPLY\n",
wu_item.id, wu_item.name, res_item.res_id, res_item.res_name,
res_item.res_report_deadline, (int)now
);
res_item.res_server_state = RESULT_SERVER_STATE_OVER;
res_item.res_outcome = RESULT_OUTCOME_NO_REPLY;
retval = transitioner.update_result(res_item);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] [RESULT#%d %s] update_result(): %d\n",
wu_item.id, wu_item.name, res_item.res_id,
res_item.res_name, retval
);
}
penalize_host(res_item.res_hostid, (double)wu_item.delay_bound);
nover++;
nno_reply++;
} else {
ninprogress++;
}
break;
case RESULT_SERVER_STATE_OVER:
nover++;
switch (res_item.res_outcome) {
case RESULT_OUTCOME_COULDNT_SEND:
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] [RESULT#%d %s] result couldn't be sent\n",
wu_item.id, wu_item.name, res_item.res_id, res_item.res_name
);
ncouldnt_send++;
break;
case RESULT_OUTCOME_SUCCESS:
if (res_item.res_validate_state == VALIDATE_STATE_INIT) {
if (canonical_result_files_deleted) {
res_item.res_validate_state = VALIDATE_STATE_TOO_LATE;
retval = transitioner.update_result(res_item);
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] [RESULT#%d %s] validate_state:INIT=>TOO_LATE retval %d\n",
wu_item.id, wu_item.name, res_item.res_id,
res_item.res_name, retval
);
} else {
have_new_result_to_validate = true;
}
}
nsuccess++;
break;
case RESULT_OUTCOME_CLIENT_ERROR:
case RESULT_OUTCOME_VALIDATE_ERROR:
nerrors++;
break;
case RESULT_OUTCOME_CLIENT_DETACHED:
case RESULT_OUTCOME_NO_REPLY:
nno_reply++;
break;
case RESULT_OUTCOME_DIDNT_NEED:
ndidnt_need++;
break;
}
break;
}
}
log_messages.printf(MSG_DEBUG,
"[WU#%d %s] %d results: unsent %d, in_progress %d, over %d (success %d, error %d, couldnt_send %d, no_reply %d, didnt_need %d)\n",
wu_item.id, wu_item.name, ntotal, nunsent, ninprogress, nover,
nsuccess, nerrors, ncouldnt_send, nno_reply, ndidnt_need
);
// if there's a new result to validate, trigger validation
//
if (have_new_result_to_validate && (nsuccess >= wu_item.min_quorum)) {
wu_item.need_validate = true;
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] need_validate:=>true\n", wu_item.id, wu_item.name
);
}
// check for WU error conditions
// NOTE: check on max # of success results is done in validater
//
if (ncouldnt_send > 0) {
wu_item.error_mask |= WU_ERROR_COULDNT_SEND_RESULT;
}
// if WU has results with errors and no success yet,
// reset homogeneous redundancy class to give other platforms a try
//
if (nerrors & !(nsuccess || ninprogress)) {
wu_item.hr_class = 0;
}
if (nerrors > wu_item.max_error_results) {
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] WU has too many errors (%d errors for %d results)\n",
wu_item.id, wu_item.name, nerrors, (int)items.size()
);
wu_item.error_mask |= WU_ERROR_TOO_MANY_ERROR_RESULTS;
}
if ((int)items.size() > wu_item.max_total_results) {
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] WU has too many total results (%d)\n",
wu_item.id, wu_item.name, (int)items.size()
);
wu_item.error_mask |= WU_ERROR_TOO_MANY_TOTAL_RESULTS;
}
// if this WU had an error, don't send any unsent results,
// and trigger assimilation if needed
//
if (wu_item.error_mask) {
for (i=0; i<items.size(); i++) {
TRANSITIONER_ITEM& res_item = items[i];
if (res_item.res_id) {
bool update_result = false;
switch(res_item.res_server_state) {
case RESULT_SERVER_STATE_UNSENT:
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] [RESULT#%d %s] server_state:UNSENT=>OVER; outcome:=>DIDNT_NEED\n",
wu_item.id, wu_item.name, res_item.res_id, res_item.res_name
);
res_item.res_server_state = RESULT_SERVER_STATE_OVER;
res_item.res_outcome = RESULT_OUTCOME_DIDNT_NEED;
update_result = true;
break;
case RESULT_SERVER_STATE_OVER:
switch (res_item.res_outcome) {
case RESULT_OUTCOME_SUCCESS:
switch(res_item.res_validate_state) {
case VALIDATE_STATE_INIT:
case VALIDATE_STATE_INCONCLUSIVE:
res_item.res_validate_state = VALIDATE_STATE_NO_CHECK;
update_result = true;
break;
}
}
}
if (update_result) {
retval = transitioner.update_result(res_item);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] [RESULT#%d %s] result.update() == %d\n",
wu_item.id, wu_item.name, res_item.res_id, res_item.res_name, retval
);
}
}
}
}
if (wu_item.assimilate_state == ASSIMILATE_INIT) {
wu_item.assimilate_state = ASSIMILATE_READY;
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] error_mask:%d assimilate_state:INIT=>READY\n",
wu_item.id, wu_item.name, wu_item.error_mask
);
}
} else if (wu_item.canonical_resultid == 0) {
// Here if no WU-level error.
// Generate new results if needed.
// NOTE: n must be signed
//
int n = wu_item.target_nresults - nunsent - ninprogress - nsuccess;
string values;
char value_buf[MAX_QUERY_LEN];
if (n > 0) {
log_messages.printf(
MSG_NORMAL,
"[WU#%d %s] Generating %d more results (%d target - %d unsent - %d in progress - %d success)\n",
wu_item.id, wu_item.name, n, wu_item.target_nresults, nunsent, ninprogress, nsuccess
);
for (j=0; j<n; j++) {
sprintf(suffix, "%d", max_result_suffix+j+1);
char rtfpath[256];
sprintf(rtfpath, "../%s", wu_item.result_template_file);
int priority_increase = 0;
if (nover && config.reliable_priority_on_over) {
priority_increase += config.reliable_priority_on_over;
} else if (nover && !nerrors && config.reliable_priority_on_over_except_error) {
priority_increase += config.reliable_priority_on_over_except_error;
}
retval = create_result_ti(
wu_item, rtfpath, suffix, key, config, value_buf, priority_increase
);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] create_result_ti() %d\n",
wu_item.id, wu_item.name, retval
);
return retval;
}
if (j==0) {
values = value_buf;
} else {
values += ",";
values += value_buf;
}
}
DB_RESULT r;
retval = r.insert_batch(values);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] insert_batch() %d\n",
wu_item.id, wu_item.name, retval
);
return retval;
}
}
}
// scan results:
// - see if all over and validated
//
all_over_and_validated = true;
bool all_over_and_ready_to_assimilate = true; // used for the defer assmilation
int most_recently_returned = 0;
for (i=0; i<items.size(); i++) {
TRANSITIONER_ITEM& res_item = items[i];
if (res_item.res_id) {
if (res_item.res_server_state == RESULT_SERVER_STATE_OVER) {
if ( res_item.res_received_time > most_recently_returned ) {
most_recently_returned = res_item.res_received_time;
}
if (res_item.res_outcome == RESULT_OUTCOME_SUCCESS) {
if (res_item.res_validate_state == VALIDATE_STATE_INIT) {
all_over_and_validated = false;
all_over_and_ready_to_assimilate = false;
}
} else if ( res_item.res_outcome == RESULT_OUTCOME_NO_REPLY ) {
if ( ( res_item.res_report_deadline + config.grace_period_hours*60*60 ) > now ) {
all_over_and_validated = false;
}
}
} else {
all_over_and_validated = false;
all_over_and_ready_to_assimilate = false;
}
}
}
// If we are defering assimilation until all results are over
// and validated then when that happens we need to make sure
// that it gets advanced to assimilate ready
// the items.size is a kludge
//
if (all_over_and_ready_to_assimilate == true && wu_item.assimilate_state == ASSIMILATE_INIT && items.size() > 0 && wu_item.canonical_resultid > 0
) {
wu_item.assimilate_state = ASSIMILATE_READY;
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] Deferred assimililation now set to ASSIMILATE_STATE_READY\n",
wu_item.id, wu_item.name
);
}
// if WU is assimilated, trigger file deletion
//
if (wu_item.assimilate_state == ASSIMILATE_DONE && ((most_recently_returned + config.delete_delay_hours*60*60) < now)) {
// can delete input files if all results OVER
//
if (all_over_and_validated && wu_item.file_delete_state == FILE_DELETE_INIT) {
wu_item.file_delete_state = FILE_DELETE_READY;
log_messages.printf(MSG_DEBUG,
"[WU#%d %s] ASSIMILATE_DONE: file_delete_state:=>READY\n",
wu_item.id, wu_item.name
);
}
// output of error results can be deleted immediately;
// output of success results can be deleted if validated
//
for (i=0; i<items.size(); i++) {
TRANSITIONER_ITEM& res_item = items[i];
// can delete canonical result outputs only if all successful
// results have been validated
//
if (((int)i == canonical_result_index) && !all_over_and_validated) {
continue;
}
if (res_item.res_id) {
do_delete = false;
switch(res_item.res_outcome) {
case RESULT_OUTCOME_CLIENT_ERROR:
do_delete = true;
break;
case RESULT_OUTCOME_SUCCESS:
do_delete = (res_item.res_validate_state != VALIDATE_STATE_INIT);
break;
}
if (do_delete && res_item.res_file_delete_state == FILE_DELETE_INIT) {
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] [RESULT#%d %s] file_delete_state:=>READY\n",
wu_item.id, wu_item.name, res_item.res_id, res_item.res_name
);
res_item.res_file_delete_state = FILE_DELETE_READY;
retval = transitioner.update_result(res_item);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] [RESULT#%d %s] result.update() == %d\n",
wu_item.id, wu_item.name, res_item.res_id, res_item.res_name, retval
);
}
}
}
}
} else if ( wu_item.assimilate_state == ASSIMILATE_DONE ) {
log_messages.printf(MSG_DEBUG,
"[WU#%d %s] not checking for items to be ready for delete because the deferred delete time has not expired. That will occur in %d seconds\n",
wu_item.id,
wu_item.name,
most_recently_returned + config.delete_delay_hours*60*60-(int)now
);
}
// compute next transition time = minimum timeout of in-progress results
//
if (wu_item.canonical_resultid) {
wu_item.transition_time = INT_MAX;
} else {
// If there is no canonical result, make sure that the transitioner will 'see'
// this WU again. In principle this is NOT needed, but it is one way to make
// the BOINC back-end more robust.
//
const int ten_days = 10*86400;
int long_delay = (int)(1.5*wu_item.delay_bound);
wu_item.transition_time = (long_delay > ten_days) ? long_delay : ten_days;
wu_item.transition_time += time(0);
}
int max_grace_or_delay_time = 0;
for (i=0; i<items.size(); i++) {
TRANSITIONER_ITEM& res_item = items[i];
if (res_item.res_id) {
if (res_item.res_server_state == RESULT_SERVER_STATE_IN_PROGRESS) {
// In cases where a result has been RESENT to a host, the
// report deadline time may be EARLIER than
// sent_time + delay_bound
// because the sent_time has been updated with the later
// "resend" time.
//
// x = res_item.res_sent_time + wu_item.delay_bound;
x = res_item.res_report_deadline;
if (x < wu_item.transition_time) {
wu_item.transition_time = x;
}
} else if ( res_item.res_server_state == RESULT_SERVER_STATE_OVER ) {
if ( res_item.res_outcome == RESULT_OUTCOME_NO_REPLY ) {
// Transition again after the grace period has expired
if ( ( res_item.res_report_deadline + config.grace_period_hours*60*60 ) > now ) {
x = res_item.res_report_deadline + config.grace_period_hours*60*60;
if (x > max_grace_or_delay_time) {
max_grace_or_delay_time = x;
}
}
} else if ( res_item.res_outcome == RESULT_OUTCOME_SUCCESS || res_item.res_outcome == RESULT_OUTCOME_CLIENT_ERROR || res_item.res_outcome == RESULT_OUTCOME_VALIDATE_ERROR) {
// Transition again after deferred delete period has experied
if ( (res_item.res_received_time + config.delete_delay_hours*60*60) > now ) {
x = res_item.res_received_time + config.delete_delay_hours*60*60;
if (x > max_grace_or_delay_time && res_item.res_received_time > 0) {
max_grace_or_delay_time = x;
}
}
}
}
}
}
// If either of the grace period or delete delay is less than
// the next transition time then use that value
//
if ( max_grace_or_delay_time < wu_item.transition_time && max_grace_or_delay_time > now && ninprogress == 0) {
wu_item.transition_time = max_grace_or_delay_time;
log_messages.printf(MSG_NORMAL,
"[WU#%d %s] Delaying transition due to grace period or delete day. New transition time = %d sec\n",
wu_item.id, wu_item.name, wu_item.transition_time
);
}
// If transition time is in the past,
// the system is bogged down and behind schedule.
// Delay processing of the WU by an amount DOUBLE the amount we are behind,
// but not less than 60 secs or more than one day.
//
if (wu_item.transition_time < now) {
int extra_delay = 2*(now - wu_item.transition_time);
if (extra_delay < 60) extra_delay = 60;
if (extra_delay > 86400) extra_delay = 86400;
log_messages.printf(MSG_DEBUG,
"[WU#%d %s] transition time in past: adding extra delay %d sec\n",
wu_item.id, wu_item.name, extra_delay
);
wu_item.transition_time = now + extra_delay;
}
log_messages.printf(MSG_DEBUG,
"[WU#%d %s] setting transition_time to %d\n",
wu_item.id, wu_item.name, wu_item.transition_time
);
retval = transitioner.update_workunit(wu_item, wu_item_original);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] workunit.update() == %d\n",
wu_item.id, wu_item.name, retval
);
return retval;
}
return 0;
}
bool do_pass() {
int retval;
DB_TRANSITIONER_ITEM_SET transitioner;
std::vector<TRANSITIONER_ITEM> items;
bool did_something = false;
if (!one_pass) check_stop_daemons();
// loop over entries that are due to be checked
//
while (1) {
retval = transitioner.enumerate(
(int)time(0), SELECT_LIMIT, mod_n, mod_i, items
);
if (retval) {
if (retval != ERR_DB_NOT_FOUND) {
log_messages.printf(MSG_CRITICAL,
"WU enum error%d; exiting\n", retval
);
exit(1);
}
break;
}
did_something = true;
TRANSITIONER_ITEM& wu_item = items[0];
retval = handle_wu(transitioner, items);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d %s] handle_wu: %d; quitting\n",
wu_item.id, wu_item.name, retval
);
exit(1);
}
if (!one_pass) check_stop_daemons();
}
return did_something;
}
void main_loop() {
int retval;
retval = boinc_db.open(config.db_name, config.db_host, config.db_user, config.db_passwd);
if (retval) {
log_messages.printf(MSG_CRITICAL, "boinc_db.open: %d\n", retval);
exit(1);
}
while (1) {
log_messages.printf(MSG_DEBUG, "doing a pass\n");
if (!do_pass()) {
if (one_pass) break;
#ifdef GCL_SIMULATOR
continue_simulation("transitioner");
signal(SIGUSR2, simulator_signal_handler);
pause();
#else
log_messages.printf(MSG_DEBUG, "sleeping %d\n", SLEEP_INTERVAL);
sleep(SLEEP_INTERVAL);
#endif
}
}
}
int main(int argc, char** argv) {
int i, retval;
char path[256];
startup_time = time(0);
for (i=1; i<argc; i++) {
if (!strcmp(argv[i], "-one_pass")) {
one_pass = true;
} else if (!strcmp(argv[i], "-d")) {
log_messages.set_debug_level(atoi(argv[++i]));
} else if (!strcmp(argv[i], "-mod")) {
mod_n = atoi(argv[++i]);
mod_i = atoi(argv[++i]);
do_mod = true;
}
}
if (!one_pass) check_stop_daemons();
retval = config.parse_file("..");
if (retval) {
log_messages.printf(MSG_CRITICAL, "can't read config file\n");
exit(1);
}
sprintf(path, "%s/upload_private", config.key_dir);
retval = read_key_file(path, key);
if (retval) {
log_messages.printf(MSG_CRITICAL, "can't read key\n");
exit(1);
}
log_messages.printf(MSG_NORMAL, "Starting\n");
install_stop_signal_handler();
main_loop();
}
const char *BOINC_RCSID_be98c91511 = "$Id$";
Reference in New Issue View Git Blame Copy Permalink