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

446 lines
14 KiB
C++
Raw Blame History

// This file is part of BOINC.
// http://boinc.berkeley.edu
// Copyright (C) 2019 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/>.
// SCHED_SHMEM is the structure of a chunk of memory shared between
// the feeder (which reads from the database)
// and instances of the scheduling server
#include "config.h"
#include <cstdio>
#include <cstdlib>
#include <cstring>
#include <string>
#include <vector>
#include <sys/param.h>
using std::vector;
#include "boinc_db.h"
#include "error_numbers.h"
#include "filesys.h"
#ifdef _USING_FCGI_
#include "boinc_fcgi.h"
#endif
#include "sched_config.h"
#include "sched_msgs.h"
#include "sched_types.h"
#include "sched_util.h"
#include "sched_shmem.h"
void SCHED_SHMEM::init(int nwu_results) {
int size = sizeof(SCHED_SHMEM) + nwu_results*sizeof(WU_RESULT);
memset(this, 0, size);
ss_size = size;
platform_size = sizeof(PLATFORM);
app_size = sizeof(APP);
app_version_size = sizeof(APP_VERSION);
assignment_size = sizeof(ASSIGNMENT);
wu_result_size = sizeof(WU_RESULT);
max_platforms = MAX_PLATFORMS;
max_apps = MAX_APPS;
max_app_versions = MAX_APP_VERSIONS;
max_assignments = MAX_ASSIGNMENTS;
max_wu_results = nwu_results;
}
static int error_return(const char* p, int expe, int got) {
fprintf(stderr, "shmem: size mismatch in %s: expected %d, got %d\n", p, expe, got);
return ERR_SCHED_SHMEM;
}
int SCHED_SHMEM::verify() {
if (platform_size != sizeof(PLATFORM)) {
return error_return("platform", sizeof(PLATFORM), platform_size);
}
if (app_size != sizeof(APP)) {
return error_return("app", sizeof(APP), app_size);
}
if (app_version_size != sizeof(APP_VERSION)) {
return error_return("app_version", sizeof(APP_VERSION), app_version_size);
}
if (assignment_size != sizeof(ASSIGNMENT)) {
return error_return("assignment", sizeof(ASSIGNMENT), assignment_size);
}
if (wu_result_size != sizeof(WU_RESULT)) {
return error_return("wu_result", sizeof(WU_RESULT), wu_result_size);
}
if (max_platforms != MAX_PLATFORMS) {
return error_return("max platform", MAX_PLATFORMS, max_platforms);
}
if (max_apps != MAX_APPS) {
return error_return("max apps", MAX_APPS, max_apps);
}
if (max_app_versions != MAX_APP_VERSIONS) {
return error_return("max app versions", MAX_APP_VERSIONS, max_app_versions);
}
if (max_assignments != MAX_ASSIGNMENTS) {
return error_return("max assignments", MAX_ASSIGNMENTS, max_assignments);
}
int size = sizeof(SCHED_SHMEM) + max_wu_results*sizeof(WU_RESULT);
if (ss_size != size) {
return error_return("shmem segment", size, ss_size);
}
return 0;
}
static void overflow(const char* table, const char* param_name) {
log_messages.printf(MSG_CRITICAL,
"The SCHED_SHMEM structure is too small for the %s table.\n"
"Either increase the %s parameter in sched_shmem.h and recompile,\n"
"or prune old rows from the table.\n"
"Then restart the project.\n",
table, param_name
);
exit(1);
}
int SCHED_SHMEM::scan_tables() {
DB_PLATFORM platform;
DB_APP app;
DB_APP_VERSION app_version;
DB_ASSIGNMENT assignment;
int i, j, n;
n = 0;
while (!platform.enumerate("where deprecated=0")) {
platforms[n++] = platform;
if (n == MAX_PLATFORMS) {
overflow("platforms", "MAX_PLATFORMS");
}
}
nplatforms = n;
n = 0;
app_weight_sum = 0;
while (!app.enumerate("where deprecated=0")) {
if (n == MAX_APPS) {
overflow("apps", "MAX_APPS");
}
app_weight_sum += app.weight;
if (app.locality_scheduling == LOCALITY_SCHED_LITE) {
locality_sched_lite = true;
}
if (app.non_cpu_intensive) {
have_nci_app = true;
}
if (config.non_cpu_intensive) {
have_nci_app = true;
app.non_cpu_intensive = true;
}
if (app.n_size_classes > 1) {
char path[MAXPATHLEN];
sprintf(path, "../size_census_%s", app.name);
#ifndef _USING_FCGI_
FILE* f = fopen(path, "r");
#else
FCGI_FILE* f = FCGI::fopen(path, "r");
#endif
if (!f) {
log_messages.printf(MSG_CRITICAL,
"Missing size census file for app %s\n", app.name
);
return ERR_FOPEN;
}
for (int i=0; i<app.n_size_classes-1; i++) {
char buf[256];
char* p = fgets(buf, 256, f);
if (!p) {
log_messages.printf(MSG_CRITICAL,
"Size census file for app %s is too short\n", app.name
);
fclose(f);
return ERR_XML_PARSE; // whatever
}
app.size_class_quantiles[i] = atof(buf);
}
fclose(f);
}
apps[n++] = app;
}
napps = n;
n = 0;
// for each (app, platform) pair,
// get all versions with numbers maximal in their plan class.
//
// If the app is using version-specific WUs,
// get all versions >= min_version,
// and flag the latest ones
//
for (i=0; i<nplatforms; i++) {
PLATFORM& splatform = platforms[i];
for (j=0; j<napps; j++) {
APP& sapp = apps[j];
vector<APP_VERSION> avs;
char query[1024];
sprintf(query,
"where appid=%lu and platformid=%lu and deprecated=0",
sapp.id, splatform.id
);
while (!app_version.enumerate(query)) {
avs.push_back(app_version);
}
// flag non-latest versions as deprecated
//
for (unsigned int k=0; k<avs.size(); k++) {
APP_VERSION& av1 = avs[k];
for (unsigned int kk=0; kk<avs.size(); kk++) {
if (k == kk) continue;
APP_VERSION& av2 = avs[kk];
if (!strcmp(av1.plan_class, av2.plan_class) && av1.version_num > av2.version_num) {
av2.deprecated = 1;
}
}
}
for (unsigned int k=0; k<avs.size(); k++) {
APP_VERSION& av1 = avs[k];
if (sapp.min_version) {
if (av1.version_num < sapp.min_version) {
fprintf(stderr, "version too small %d < %d\n",
av1.version_num, sapp.min_version
);
continue;
}
} else {
if (av1.deprecated) continue;
}
if (av1.min_core_version && av1.min_core_version < 10000) {
fprintf(stderr, "min core version too small - multiplying by 100\n");
av1.min_core_version *= 100;
}
if (av1.max_core_version && av1.max_core_version < 10000) {
fprintf(stderr, "max core version too small - multiplying by 100\n");
av1.max_core_version *= 100;
}
app_versions[n++] = av1;
if (n == MAX_APP_VERSIONS) {
overflow("app_versions", "MAX_APP_VERSIONS");
}
}
}
}
napp_versions = n;
// see which resources we have app versions for
//
for (i=0; i<NPROC_TYPES; i++) {
have_apps_for_proc_type[i] = false;
}
for (i=0; i<napp_versions; i++) {
APP_VERSION& av = app_versions[i];
int rt = plan_class_to_proc_type(av.plan_class);
have_apps_for_proc_type[rt] = true;
}
for (i=0; i<NPROC_TYPES; i++) {
fprintf(stderr, "have apps for %s: %s\n",
proc_type_name(i),
have_apps_for_proc_type[i]?"yes":"no"
);
}
n = 0;
while (!assignment.enumerate("where multi <> 0")) {
assignments[n++] = assignment;
if (n == MAX_ASSIGNMENTS) {
overflow("assignments", "MAX_ASSIGNMENTS");
}
}
nassignments = n;
return 0;
}
PLATFORM* SCHED_SHMEM::lookup_platform(char* name) {
for (int i=0; i<nplatforms; i++) {
if (!strcmp(platforms[i].name, name)) {
return &platforms[i];
}
}
return NULL;
}
PLATFORM* SCHED_SHMEM::lookup_platform_id(DB_ID_TYPE id) {
for (int i=0; i<nplatforms; i++) {
if (platforms[i].id == id) return &platforms[i];
}
return NULL;
}
APP* SCHED_SHMEM::lookup_app(DB_ID_TYPE id) {
for (int i=0; i<napps; i++) {
if (apps[i].id == id) return &apps[i];
}
return NULL;
}
APP* SCHED_SHMEM::lookup_app_name(char* name) {
for (int i=0; i<napps; i++) {
if (!strcmp(name, apps[i].name)) return &apps[i];
}
return NULL;
}
APP_VERSION* SCHED_SHMEM::lookup_app_version(DB_ID_TYPE id) {
APP_VERSION* avp;
for (int i=0; i<napp_versions; i++) {
avp = &app_versions[i];
if (avp->id == id) {
return avp;
}
}
return NULL;
}
APP_VERSION* SCHED_SHMEM::lookup_app_version_platform_plan_class(
int platformid, char* plan_class
) {
APP_VERSION* avp;
for (int i=0; i<napp_versions; i++) {
avp = &app_versions[i];
if (avp->platformid == platformid && !strcmp(avp->plan_class, plan_class)) {
return avp;
}
}
return NULL;
}
// see if there's any work.
// If there is, reserve it for this process
// (if we don't do this, there's a race condition where lots
// of servers try to get a single work item)
//
bool SCHED_SHMEM::no_work(int pid) {
if (!ready) return true;
for (int i=0; i<max_wu_results; i++) {
if (wu_results[i].state == WR_STATE_PRESENT) {
wu_results[i].state = pid;
return false;
}
}
return true;
}
void SCHED_SHMEM::restore_work(int pid) {
for (int i=0; i<max_wu_results; i++) {
if (wu_results[i].state == pid) {
wu_results[i].state = WR_STATE_PRESENT;
return;
}
}
}
void SCHED_SHMEM::show(FILE* f) {
fprintf(f, "apps:\n");
for (int i=0; i<napps; i++) {
APP& app = apps[i];
fprintf(f, "id: %lu name: %s hr: %d weight: %.2f beta: %d hav: %d nci: %d\n",
app.id, app.name, app.homogeneous_redundancy, app.weight,
app.beta, app.homogeneous_app_version, app.non_cpu_intensive
);
}
fprintf(f, "app versions:\n");
for (int i=0; i<napp_versions; i++) {
APP_VERSION av = app_versions[i];
fprintf(f, "appid: %lu platformid: %lu version_num: %d plan_class: %s\n",
av.appid, av.platformid, av.version_num, av.plan_class
);
}
for (int i=0; i<NPROC_TYPES; i++) {
fprintf(f,
"have %s apps: %s\n",
proc_type_name(i),
have_apps_for_proc_type[i]?"yes":"no"
);
}
fprintf(f,
"Jobs; key:\n"
"ap: app ID\n"
"ic: infeasible count\n"
"wu: workunit ID\n"
"rs: result ID\n"
"hr: HR class\n"
"nr: need reliable\n"
);
fprintf(f,
"host fpops mean %f stddev %f\n",
perf_info.host_fpops_mean, perf_info.host_fpops_stddev
);
fprintf(f,
"host fpops 50th pctile %f 95th pctile %f\n",
perf_info.host_fpops_50_percentile, perf_info.host_fpops_95_percentile
);
fprintf(f, "ready: %d\n", ready);
fprintf(f, "max_wu_results: %d\n", max_wu_results);
for (int i=0; i<max_wu_results; i++) {
if (i%24 == 0) {
fprintf(f,
"%4s %12s %10s %10s %10s %8s %10s %8s %12s %12s %9s %9s\n",
"slot",
"app",
"WU ID",
"result ID",
"batch",
"HR class",
"priority",
"in shmem",
"size class",
"need reliable",
"inf count",
"version"
);
}
WU_RESULT& wu_result = wu_results[i];
APP* app;
const char* appname;
int delta_t;
switch(wu_result.state) {
case WR_STATE_PRESENT:
app = lookup_app(wu_result.workunit.appid);
appname = app?app->name:"missing";
delta_t = dtime() - wu_result.time_added_to_shared_memory;
fprintf(f,
"%4d %12.12s %10lu %10lu %10d %8d %10d %7ds %9d %12s %9d %9d\n",
i,
appname,
wu_result.workunit.id,
wu_result.resultid,
wu_result.workunit.batch,
wu_result.workunit.hr_class,
wu_result.res_priority,
delta_t,
wu_result.workunit.size_class,
wu_result.need_reliable?"yes":"no",
wu_result.infeasible_count,
wu_result.workunit.app_version_num
);
break;
case WR_STATE_EMPTY:
fprintf(f, "%4d: ---\n", i);
break;
default:
fprintf(f, "%4d: PID %d: result %lu\n", i, wu_result.state, wu_result.resultid);
}
}
}
Reference in New Issue View Git Blame Copy Permalink