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

605 lines
20 KiB
C
Raw Blame History

// Berkeley Open Infrastructure for Network Computing
// http://boinc.berkeley.edu
// Copyright (C) 2005 University of California
//
// This 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 2.1 of the License, or (at your option) any later version.
//
// This software 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.
//
// To view the GNU Lesser General Public License visit
// http://www.gnu.org/copyleft/lesser.html
// or write to the Free Software Foundation, Inc.,
// 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
// -------------------------------
//
// feeder
// [ -asynch ] fork and run in a separate process
// [ -d x ] debug level x
// [ -random_order ] order by "random" field of result
// [ -priority_order ] order by decreasing "priority" field of result
// [ -priority_order_create_time ]
// order by priority, then by increasing WU create time
// [ -mod n i ] handle only results with (id mod n) == i
// [ -sleep_interval x ] sleep x seconds if nothing to do
// [ -allapps ] interleave results from all applications uniformly
// [ -purge_stale x ] remove work items from the shared memory segment
// that have been there for longer then x minutes
// but haven't been assigned
// [ -reliable x ] flag results for workunits older then x days as "need_reliable"
//
// Creates a shared memory segment containing DB info,
// including the work array (results/workunits to send).
//
// Try to keep the work array filled.
// This is a little tricky.
// We use an enumerator.
// The inner loop scans the wu_result table,
// looking for empty slots and trying to fill them in.
// When the enumerator reaches the end, it is restarted;
// hopefully there will be some new workunits.
// There are two complications:
//
// - An enumeration may return results already in the array.
// So, for each result, we scan the entire array to make sure
// it's not there already. Can this be streamlined?
//
// - We must avoid excessive re-enumeration,
// especially when the number of results is less than the array size.
// Crude approach: if a "collision" (as above) occurred on
// a pass through the array, wait a long time (5 sec)
//
// Checking for infeasible results (i.e. can't sent to any host):
//
// - the "infeasible_count" field of WU_RESULT keeps track of
// how many times the WU_RESULT was infeasible for a host
//
// - the scheduler gives priority to results that have infeasible_count > 0
//
// - If the infeasible_count of any result exceeds MAX_INFEASIBLE_COUNT,
// the feeder flags the result as OVER with outcome COULDNT_SEND,
// and flags the WU for the transitioner.
//
// - the feeder tries to ensure that the number of WU_RESULTs
// with infeasible_count > MAX_INFEASIBLE_THRESHOLD
// doesn't exceed MAX_INFEASIBLE (defined in sched_shmem.h)
// If it does, then the feeder picks the WU_RESULT with
// the largest infeasible_count, marks if COULDNT_SEND as above,
// and repeats this until the infeasible count is low enough again
// Trigger files:
// The feeder program periodically checks for two trigger files:
//
// stop_server: destroy shmem and exit
// leave trigger file there (for other daemons)
// reread_db: update DB contents in existing shmem
// delete trigger file
// If you get an "Invalid argument" error when trying to run the feeder,
// it is likely that you aren't able to allocate enough shared memory.
// Either increase the maximum shared memory segment size in the kernel
// configuration, or decrease the MAX_PLATFORMS, MAX_APPS
// MAX_APP_VERSIONS, and MAX_WU_RESULTS in sched_shmem.h
#include "config.h"
#include <cstdio>
#include <cstdlib>
#include <cstring>
#include <ctime>
#include <unistd.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <vector>
using std::vector;
#include "boinc_db.h"
#include "shmem.h"
#include "error_numbers.h"
#include "synch.h"
#include "util.h"
#include "sched_config.h"
#include "sched_shmem.h"
#include "sched_util.h"
#include "sched_msgs.h"
#define DEFAULT_SLEEP_INTERVAL 5
#define ENUM_LIMIT MAX_WU_RESULTS*2
// The following parameters determine the feeder's policy
// for purging "infeasible" results,
// i.e. those that are hard to send to any client.
// TODO: remove these from the source code,
// make them config.xml parameters
#define MAX_INFEASIBLE_THRESHOLD 2000
// if a result's infeasible_count exceeds this,
// count it as "possibly infeasible" (see the following)
// TODO: lower this to 20 or so
#define MAX_INFEASIBLE 500
// if # of possibly infeasibly results exceeds this,
// classify some of them as COULDNT_SEND and remove from array
#define MAX_INFEASIBLE_COUNT 5000
// a result's infeasible_count exceeds this,
// classify as COULDNT_SEND and remove it from array
// TODO: lower this to 50 or so
// Uncomment the following to enable this purging.
//
//#define REMOVE_INFEASIBLE_ENTRIES
#define REREAD_DB_FILENAME "../reread_db"
SCHED_CONFIG config;
SCHED_SHMEM* ssp;
key_t sema_key;
const char* order_clause="";
char select_clause[256];
double sleep_interval = DEFAULT_SLEEP_INTERVAL;
bool all_apps = false;
int purge_stale_time = 0;
int reliable_time = 0;
void cleanup_shmem() {
ssp->ready = false;
detach_shmem((void*)ssp);
destroy_shmem(config.shmem_key);
}
int check_reread_trigger() {
FILE* f;
f = fopen(REREAD_DB_FILENAME, "r");
if (f) {
fclose(f);
log_messages.printf(
SCHED_MSG_LOG::MSG_NORMAL,
"Found trigger file %s; re-scanning database tables.\n",
REREAD_DB_FILENAME
);
ssp->init();
ssp->scan_tables();
unlink(REREAD_DB_FILENAME);
log_messages.printf(
SCHED_MSG_LOG::MSG_NORMAL,
"Done re-scanning: trigger file removed.\n"
);
}
return 0;
}
#ifdef REMOVE_INFEASIBLE_ENTRIES
static int remove_infeasible(int i) {
char buf[256];
int retval;
DB_RESULT result;
DB_WORKUNIT wu;
WU_RESULT& wu_result = ssp->wu_results[i];
wu_result.state = WR_STATE_EMPTY;
result = wu_result.result;
wu = wu_result.workunit;
log_messages.printf(
SCHED_MSG_LOG::MSG_NORMAL,
"[%s] declaring result as unsendable; infeasible count %d\n",
result.name, wu_result.infeasible_count
);
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) {
log_messages.printf(
SCHED_MSG_LOG::MSG_CRITICAL,
"[%s]: can't update: %d\n",
result.name, retval
);
return retval;
}
wu.transition_time = time(0);
sprintf(buf, "transition_time=%d", wu.transition_time);
retval = wu.update_field(buf);
if (retval) {
log_messages.printf(
SCHED_MSG_LOG::MSG_CRITICAL,
"[%s]: can't update: %d\n",
wu.name, retval
);
return retval;
}
return 0;
}
#endif
// Scan work items for a given appp until one found
// that is not already on the shared memory segment.
// Errors that can occur:
// 1) No valid work item found even after restarting the enumeration
// ACTION: return false
// 2) The work item can be for a app that doesn't exist in the database
// ACTION: exit application
//
static bool find_work_item(
DB_WORK_ITEM *wi, bool& restarted_enum, int& ncollisions,
int work_item_index, int enum_size, char* mod_select_clause
) {
bool in_second_pass = false;
bool found = false;
int retval, j;
if (!wi->cursor.active && restarted_enum) {
return false;
} else if (!wi->cursor.active) {
in_second_pass = true;
}
do {
// if we have restarted the enum then we are in the second pass
//
if (!in_second_pass && restarted_enum ) {
in_second_pass = true;
ncollisions = 0;
}
retval = wi->enumerate(enum_size, mod_select_clause, order_clause);
// if retval is not 0 (i.e. true),
// then we have reached the end of the result
// and we need to requery the database
//
if (retval) {
restarted_enum = true;
log_messages.printf(SCHED_MSG_LOG::MSG_NORMAL,
"restarted enumeration for appid %d\n",
ssp->apps[work_item_index].id);
} else {
// Check for a work item with an invalid application id
//
if (!ssp->have_app(wi->wu.appid)) {
log_messages.printf(
SCHED_MSG_LOG::MSG_CRITICAL,
"result [RESULT#%d] has bad appid %d; clean up your DB!\n",
wi->res_id, wi->wu.appid
);
exit(1);
}
// Check for collision
// (i.e. this result already is in the array)
// If collision, then advance to the next workitem
//
found = true;
for (j=0; j<ssp->nwu_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++;
found = false; // not found if it is a collision
break;
}
}
}
// exit conditions
// (in_second_pass && retval) means if we have looped a second time
// and reached the end of the result set without finding a workitem.
// This is an error.
// found means that we identified a valid work item
//
} while ((!in_second_pass || !retval) && !found);
return found;
}
static int find_work_item_index(int slot_pos) {
if (ssp->app_weights == 0) return 0;
int mod = slot_pos % (int)(ssp->app_weights);
int work_item_index = -1;
for (int i=0; i<ssp->napps; i++) {
if (ssp->apps[i].weight < 1) continue;
if (mod < ssp->apps[i].weight) {
work_item_index = i;
break;
} else {
mod = mod - (int)ssp->apps[i].weight;
}
}
// The condition below will occur if all projects have a weight of 0
if ( work_item_index == -1 ) work_item_index = 0;
return work_item_index;
}
static void scan_work_array(
vector<DB_WORK_ITEM>* work_items,
int& nadditions, int& ncollisions, int& /*ninfeasible*/
) {
int i;
bool found;
bool restarted_enum[ssp->napps];
DB_WORK_ITEM* wi;
char mod_select_clause[256];
int work_item_index;
int enum_size;
for (i=0; i < ssp->napps; i++) {
restarted_enum[i] = false;
}
for (i=0; i<ssp->nwu_results; i++) {
// If all_apps is set then every nth item in the shared memory segment
// will be assigned to the application stored in that index in ssp->apps
//
if (all_apps) {
if (ssp->app_weights > 0) {
work_item_index = find_work_item_index(i);
enum_size = (int) floor(0.5 + ENUM_LIMIT*(ssp->apps[work_item_index].weight)/(ssp->app_weights));
} else {
// If all apps have a weight of zero then evenly distribute the slots
work_item_index = i % ssp->napps;
enum_size = (int) floor(0.5 + ((double)ENUM_LIMIT)/ssp->napps);
}
sprintf(mod_select_clause, "%s and result.appid=%d",
select_clause, ssp->apps[work_item_index].id
);
} else {
work_item_index = 0;
enum_size = ENUM_LIMIT;
strcpy(mod_select_clause, select_clause);
}
wi = &((*work_items).at(work_item_index));
WU_RESULT& wu_result = ssp->wu_results[i];
switch (wu_result.state) {
case WR_STATE_PRESENT:
if (purge_stale_time && wu_result.time_added_to_shared_memory < (time(0) - purge_stale_time)) {
wu_result.state = WR_STATE_EMPTY;
log_messages.printf(SCHED_MSG_LOG::MSG_NORMAL,
"remove result [RESULT#%d] from slot %d because it is stale\n",
wu_result.resultid, i);
} else {
break;
}
#ifdef REMOVE_INFEASIBLE_ENTRIES
if (wu_result.infeasible_count > MAX_INFEASIBLE_COUNT) {
remove_infeasible(i);
} else if (wu_result.infeasible_count > MAX_INFEASIBLE_THRESHOLD) {
ninfeasible++;
}
break;
#endif
case WR_STATE_EMPTY:
found = find_work_item(
wi, restarted_enum[work_item_index], ncollisions,
work_item_index, enum_size, mod_select_clause
);
if (found) {
log_messages.printf(
SCHED_MSG_LOG::MSG_NORMAL,
"adding result [RESULT#%d] in slot %d\n",
wi->res_id, i
);
wu_result.resultid = wi->res_id;
wu_result.workunit = wi->wu;
wu_result.state = WR_STATE_PRESENT;
// If the workunit has already been allocated to a certain
// OS then it should be assigned quickly,
// so we set its infeasible_count to 1
if (wi->wu.hr_class > 0) {
wu_result.infeasible_count = 1;
} else {
wu_result.infeasible_count = 0;
}
// If using the reliable mechanism, then set the results for
// workunits older then the specificed time as needing a reliable
// host
wu_result.need_reliable = 0;
if (reliable_time) {
if ((wu_result.workunit.create_time + reliable_time*86400) <= time(0)) {
wu_result.need_reliable = true;
}
}
wu_result.time_added_to_shared_memory = time(NULL);
nadditions++;
}
break;
default:
// here the state is a PID; see if it's still alive
//
int pid = wu_result.state;
struct stat s;
char buf[256];
sprintf(buf, "/proc/%d", pid);
if (stat(buf, &s)) {
wu_result.state = WR_STATE_PRESENT;
log_messages.printf(
SCHED_MSG_LOG::MSG_NORMAL,
"Result reserved by non-existent process PID %d; resetting\n",
pid
);
}
}
}
}
#ifdef REMOVE_INFEASIBLE_ENTRIES
static int remove_most_infeasible() {
int i, max, imax=-1;
max = 0;
for (i=0; i<ssp->nwu_results; i++) {
WU_RESULT& wu_result = ssp->wu_results[i];
if (wu_result.state == WR_STATE_PRESENT && wu_result.infeasible_count > max) {
imax = i;
max = wu_result.infeasible_count;
}
}
if (max == 0) return -1; // nothing is infeasible
return remove_infeasible(imax);
}
#endif
void feeder_loop() {
int nadditions, ncollisions, ninfeasible;
vector<DB_WORK_ITEM> work_items;
DB_WORK_ITEM* wi;
if (all_apps) {
for(int i=0; i<ssp->napps; i++) {
wi = new DB_WORK_ITEM();
work_items.push_back(*wi);
}
} else {
wi = new DB_WORK_ITEM();
work_items.push_back(*wi);
}
while (1) {
nadditions = 0;
ncollisions = 0;
ninfeasible = 0;
scan_work_array(&work_items, nadditions, ncollisions, ninfeasible);
ssp->ready = true;
#ifdef REMOVE_INFEASIBLE_ENTRIES
int i, n, retval;
if (ninfeasible > MAX_INFEASIBLE) {
n = ninfeasible - MAX_INFEASIBLE;
for (i=0; i<n; i++ ) {
retval = remove_most_infeasible();
if (retval) break;
}
}
#endif
if (nadditions == 0) {
log_messages.printf(SCHED_MSG_LOG::MSG_DEBUG, "No results added; sleeping %.2f sec\n", sleep_interval);
boinc_sleep(sleep_interval);
} else {
log_messages.printf(SCHED_MSG_LOG::MSG_DEBUG, "Added %d results to array\n", nadditions);
}
if (ncollisions) {
log_messages.printf(SCHED_MSG_LOG::MSG_DEBUG, "Some results already in array - sleeping %.2f sec\n", sleep_interval);
boinc_sleep(sleep_interval);
}
fflush(stdout);
check_stop_daemons();
check_reread_trigger();
}
}
int main(int argc, char** argv) {
int i, retval;
bool asynch = false;
void* p;
char path[256];
unlink(REREAD_DB_FILENAME);
retval = config.parse_file("..");
if (retval) {
log_messages.printf(SCHED_MSG_LOG::MSG_CRITICAL, "can't parse config file\n");
exit(1);
}
for (i=1; i<argc; i++) {
if (!strcmp(argv[i], "-asynch")) {
asynch = true;
} else if (!strcmp(argv[i], "-d")) {
log_messages.set_debug_level(atoi(argv[++i]));
} else if (!strcmp(argv[i], "-random_order")) {
order_clause = "order by random ";
} else if (!strcmp(argv[i], "-allapps")) {
all_apps = true;
} else if (!strcmp(argv[i], "-priority_order")) {
order_clause = "order by priority desc ";
} else if (!strcmp(argv[i], "-priority_order_create_time")) {
order_clause = "order by priority desc, workunit.create_time ";
} else if (!strcmp(argv[i], "-purge_stale")) {
purge_stale_time = atoi(argv[++i])*60;
} else if (!strcmp(argv[i], "-reliable")) {
reliable_time = atoi(argv[++i]);
} else if (!strcmp(argv[i], "-mod")) {
int n = atoi(argv[++i]);
int j = atoi(argv[++i]);
sprintf(select_clause, "and result.id %% %d = %d ", n, j);
} else if (!strcmp(argv[i], "-sleep_interval")) {
sleep_interval = atof(argv[++i]);
} else {
log_messages.printf(SCHED_MSG_LOG::MSG_CRITICAL,
"bad cmdline arg: %s\n", argv[i]
);
exit(1);
}
}
if (asynch) {
if (fork()!=0) {
exit(0);
}
}
log_messages.printf(SCHED_MSG_LOG::MSG_NORMAL, "Starting\n");
get_project_dir(path, sizeof(path));
get_key(path, 'a', sema_key);
destroy_semaphore(sema_key);
create_semaphore(sema_key);
retval = destroy_shmem(config.shmem_key);
if (retval) {
log_messages.printf(SCHED_MSG_LOG::MSG_CRITICAL, "can't destroy shmem\n");
exit(1);
}
retval = create_shmem(config.shmem_key, sizeof(SCHED_SHMEM), 0 /* don't set GID */, &p);
if (retval) {
log_messages.printf(SCHED_MSG_LOG::MSG_CRITICAL, "can't create shmem\n");
exit(1);
}
ssp = (SCHED_SHMEM*)p;
ssp->init();
atexit(cleanup_shmem);
install_stop_signal_handler();
retval = boinc_db.open(
config.db_name, config.db_host, config.db_user, config.db_passwd
);
if (retval) {
log_messages.printf(SCHED_MSG_LOG::MSG_CRITICAL,
"boinc_db.open: %d; %s\n", retval, boinc_db.error_string()
);
exit(1);
}
ssp->scan_tables();
log_messages.printf(
SCHED_MSG_LOG::MSG_NORMAL,
"feeder: read "
"%d platforms, "
"%d apps, "
"%d app_versions\n",
ssp->nplatforms,
ssp->napps,
ssp->napp_versions
);
feeder_loop();
}
const char *BOINC_RCSID_57c87aa242 = "$Id$";
Reference in New Issue View Git Blame Copy Permalink