// Called when there's evidence that the host has detached.
// Mark in-progress results for the given host
// as server state OVER, outcome CLIENT_DETACHED.
// This serves two purposes:
// 1) make sure we don't resend these results to the host
// (they may be the reason the user detached)
// 2) trigger the generation of new results for these WUs
//
static void mark_results_over(DB_HOST& host) {
char buf[256], buf2[256];
DB_RESULT result;
sprintf(buf, "where hostid=%d and server_state=%d",
host.id,
RESULT_SERVER_STATE_IN_PROGRESS
);
while (!result.enumerate(buf)) {
sprintf(buf2,
"server_state=%d, outcome=%d, received_time = %ld",
RESULT_SERVER_STATE_OVER,
RESULT_OUTCOME_CLIENT_DETACHED,
time(0)
);
result.update_field(buf2);
// and trigger WU transition
//
DB_WORKUNIT wu;
wu.id = result.workunitid;
sprintf(buf2, "transition_time=%d", (int)time(0));
wu.update_field(buf2);
log_messages.printf(MSG_CRITICAL,
"[HOST#%d] [RESULT#%u] [WU#%u] changed CPID: marking in-progress result %s as client error!\n",
host.id, result.id, result.workunitid, result.name
);
}
}
int handle_result(DB_RESULT& result) {
DB_WORKUNIT wu;
int retval;
char path[256];
char buf[256];
FILE* f;
retval = wu.lookup_id(result.workunitid);
if (retval) {
printf(
"ERROR: can't find WU %d for result %d\n",
result.workunitid, result.id
);
return 1;
}
get_file_path(wu, path);
f = fopen(path, "r");
if (f) {
fclose(f);
} else {
printf("no file %s for result %d\n",
path, result.id
);
if (repair) {
if (result.server_state == RESULT_SERVER_STATE_UNSENT) {
result.server_state = RESULT_SERVER_STATE_OVER;
result.outcome = RESULT_OUTCOME_COULDNT_SEND;
sprintf(
buf,"server_state=%d, outcome=%d",
result.server_state, result.outcome
);
retval = result.update_field(buf);
if (retval) {
printf(
"ERROR: can't update result %d\n",
result.id
);
return 1;
}
}
}
return 1;
}
return 0;
}
// Arrange that further results for this workunit
// will be sent only to hosts with the given user ID.
// This could be used, for example, so that late workunits
// are sent only to cloud or cluster resources
//
int restrict_wu_to_user(WORKUNIT& _wu, int userid) {
DB_RESULT result;
DB_ASSIGNMENT asg;
DB_WORKUNIT wu;
wu = _wu;
char buf[256];
int retval;
// mark unsent results as DIDNT_NEED
//
sprintf(buf, "where workunitid=%d and server_state=%d",
wu.id, RESULT_SERVER_STATE_UNSENT
);
while (!result.enumerate(buf)) {
char buf2[256];
sprintf(buf2, "server_state=%d, outcome=%d",
RESULT_SERVER_STATE_OVER,
RESULT_OUTCOME_DIDNT_NEED
);
result.update_field(buf2);
}
// mark the WU as TRANSITION_NO_NEW_RESULTS
//
sprintf(buf, "transitioner_flags=%d", TRANSITION_NO_NEW_RESULTS);
retval = wu.update_field(buf);
if (retval) return retval;
// create an assignment record
//
asg.clear();
asg.create_time = time(0);
asg.target_id = userid;
asg.target_type = ASSIGN_USER;
asg.multi = 0;
asg.workunitid = wu.id;
retval = asg.insert();
return retval;
}
// return true if we changed the file_delete_state of a WU or a result
//
bool do_pass(bool retry_error) {
DB_WORKUNIT wu;
DB_RESULT result;
bool did_something = false;
char buf[256];
char clause[256];
int retval, new_state;
check_stop_daemons();
strcpy(clause, "");
if (id_modulus) {
sprintf(clause, " and id %% %d = %d ", id_modulus, id_remainder);
}
if (dont_delete_batches) {
strcat(clause, " and batch <= 0 ");
}
if (appid) {
sprintf(buf, " and appid = %d ", appid);
strcat(clause, buf);
}
sprintf(buf,
"where file_delete_state=%d %s limit %d",
retry_error?FILE_DELETE_ERROR:FILE_DELETE_READY,
clause, WUS_PER_ENUM
);
while (do_input_files) {
retval = wu.enumerate(buf);
if (retval) {
if (retval != ERR_DB_NOT_FOUND) {
log_messages.printf(MSG_DEBUG, "DB connection lost, exiting\n");
exit(0);
}
break;
}
if (preserve_wu_files) {
retval = 0;
} else {
retval = wu_delete_files(wu);
}
if (retval) {
new_state = FILE_DELETE_ERROR;
log_messages.printf(MSG_CRITICAL,
"[WU#%d] file deletion failed: %s\n", wu.id, boincerror(retval)
);
} else {
new_state = FILE_DELETE_DONE;
}
if (new_state != wu.file_delete_state) {
sprintf(buf, "file_delete_state=%d", new_state);
retval = wu.update_field(buf);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[WU#%d] update failed: %s\n", wu.id, boincerror(retval)
);
} else {
log_messages.printf(MSG_DEBUG,
"[WU#%d] file_delete_state updated\n", wu.id
);
did_something = true;
}
}
}
sprintf(buf,
"where file_delete_state=%d %s limit %d",
retry_error?FILE_DELETE_ERROR:FILE_DELETE_READY,
clause, RESULTS_PER_ENUM
);
while (do_output_files) {
retval = result.enumerate(buf);
if (retval) {
if (retval != ERR_DB_NOT_FOUND) {
log_messages.printf(MSG_DEBUG, "DB connection lost, exiting\n");
exit(0);
}
break;
}
if (preserve_result_files) {
retval = 0;
} else {
retval = result_delete_files(result);
}
if (retval) {
new_state = FILE_DELETE_ERROR;
log_messages.printf(MSG_CRITICAL,
"[RESULT#%d] file deletion failed: %s\n", result.id, boincerror(retval)
);
} else {
new_state = FILE_DELETE_DONE;
}
if (new_state != result.file_delete_state) {
sprintf(buf, "file_delete_state=%d", new_state);
retval = result.update_field(buf);
if (retval) {
log_messages.printf(MSG_CRITICAL,
"[RESULT#%d] update failed: %s\n", result.id, boincerror(retval)
);
} else {
log_messages.printf(MSG_DEBUG,
"[RESULT#%d] file_delete_state updated\n", result.id
);
did_something = true;
}
}
}
return did_something;
}
// Enumerate jobs from DB until find one that is not already in the work array.
// If find one, return true.
// If reach end of enum for second time on this array scan, return false
//
static bool get_job_from_db(
DB_WORK_ITEM& wi, // enumerator to get job from
int app_index, // if using --allapps, the app index
int& enum_phase,
int& ncollisions
) {
bool collision;
int retval, j, enum_size;
char select_clause[256];
if (all_apps) {
sprintf(select_clause, "%s and r1.appid=%lu",
mod_select_clause, ssp->apps[app_index].id
);
enum_size = enum_sizes[app_index];
} else {
safe_strcpy(select_clause, mod_select_clause);
enum_size = enum_limit;
}
int hrt = ssp->apps[app_index].homogeneous_redundancy;
while (1) {
if (hrt && config.hr_allocate_slots) {
retval = wi.enumerate_all(enum_size, select_clause);
} else {
retval = wi.enumerate(enum_size, select_clause, order_clause);
}
if (retval) {
if (retval != ERR_DB_NOT_FOUND) {
// If DB server dies, exit;
// so /start (run from crontab) will restart us eventually.
//
log_messages.printf(MSG_CRITICAL,
"DB connection lost, exiting\n"
);
exit(0);
}
// we've reach the end of the result set
//
switch (enum_phase) {
case ENUM_FIRST_PASS:
enum_phase = ENUM_SECOND_PASS;
ncollisions = 0;
// disregard collisions - maybe we'll find new jobs
break;
case ENUM_SECOND_PASS:
enum_phase = ENUM_OVER;
return false;
}
log_messages.printf(MSG_NORMAL,
"restarted enumeration for appid %lu\n",
ssp->apps[app_index].id
);
} else {
// Check for invalid application ID
//
if (!ssp->lookup_app(wi.wu.appid)) {
#if 0
log_messages.printf(MSG_CRITICAL,
"result [RESULT#%u] has bad appid %d; clean up your DB!\n",
wi.res_id, wi.wu.appid
);
#endif
continue;
}
// if the WU had an error, mark result as DIDNT_NEED
//
if (wi.wu.error_mask) {
char buf[256];
DB_RESULT result;
result.id = wi.res_id;
sprintf(buf, "server_state=%d, outcome=%d",
RESULT_SERVER_STATE_OVER,
RESULT_OUTCOME_DIDNT_NEED
);
result.update_field(buf);
log_messages.printf(MSG_NORMAL,
"[RESULT#%lu] WU had error, marking as DIDNT_NEED\n",
wi.res_id
);
continue;
}
// Check for collision (i.e. this result already is in the array)
//
collision = false;
for (j=0; j<ssp->max_wu_results; j++) {
if (ssp->wu_results[j].state != WR_STATE_EMPTY && ssp->wu_results[j].resultid == wi.res_id) {
// If the result is already in shared mem,
// and another instance of the WU has been sent,
// bump the infeasible count to encourage
// it to get sent more quickly
//
if (ssp->wu_results[j].infeasible_count == 0) {
if (wi.wu.hr_class > 0) {
ssp->wu_results[j].infeasible_count++;
}
}
ncollisions++;
collision = true;
log_messages.printf(MSG_DEBUG,
"result [RESULT#%lu] already in array\n", wi.res_id
);
break;
}
}
if (collision) {
continue;
}
// if using HR, check whether we've exceeded quota for this class
//
if (hrt && config.hr_allocate_slots) {
if (!hr_info.accept(hrt, wi.wu.hr_class)) {
log_messages.printf(MSG_DEBUG,
"rejecting [RESULT#%lu] because HR class %d/%d over quota\n",
wi.res_id, hrt, wi.wu.hr_class
);
continue;
}
}
return true;
}
}
return false; // never reached
}
