mirror of https://github.com/BOINC/boinc.git
219 lines
7.8 KiB
C++
219 lines
7.8 KiB
C++
// 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/>.
|
|
|
|
#ifndef _SCHED_CONFIG_
|
|
#define _SCHED_CONFIG_
|
|
|
|
#include <regex.h>
|
|
#include <vector>
|
|
#include <cstdio>
|
|
|
|
#include "sched_limit.h"
|
|
|
|
using std::vector;
|
|
|
|
// constants related to consecutive_valid.
|
|
// These could be made configurable.
|
|
|
|
#define CONS_VALID_RELIABLE 10
|
|
// host is eligible to be considered "reliable"
|
|
#define CONS_VALID_HOST_SCALE 10
|
|
// host is eligible for host scaling of credit
|
|
#define CONS_VALID_UNREPLICATED 10
|
|
// host is eligible for single replication
|
|
|
|
// server configuration
|
|
// MISNAMED - should be SERVER_CONFIG,
|
|
// and should factor out scheduler-specific stuff into SCHED_CONFIG
|
|
//
|
|
struct SCHED_CONFIG {
|
|
char master_url[256];
|
|
char long_name[256];
|
|
char db_name[256];
|
|
char db_user[256];
|
|
char db_passwd[256];
|
|
char db_host[256];
|
|
char replica_db_name[256];
|
|
char replica_db_user[256];
|
|
char replica_db_passwd[256];
|
|
char replica_db_host[256];
|
|
int shmem_key;
|
|
char project_dir[256];
|
|
char key_dir[256];
|
|
char download_url[256];
|
|
char download_dir[256];
|
|
char upload_url[256];
|
|
char upload_dir[256];
|
|
int report_grace_period;
|
|
// grace period for reporting results;
|
|
// server's report deadline is client's deadline + this
|
|
double delete_delay;
|
|
bool msg_to_host;
|
|
bool non_cpu_intensive;
|
|
bool verify_files_on_app_start;
|
|
int homogeneous_redundancy;
|
|
bool hr_allocate_slots;
|
|
bool ignore_upload_certificates;
|
|
bool dont_generate_upload_certificates;
|
|
int uldl_dir_fanout; // fanout of ul/dl dirs; 0 if none
|
|
bool cache_md5_info;
|
|
int fuh_debug_level;
|
|
int reliable_priority_on_over;
|
|
// additional results generated after at least one result
|
|
// is over will have their priority boosted by this amount
|
|
int reliable_priority_on_over_except_error;
|
|
// additional results generated after at least one result is over
|
|
// (unless there is an error) will have their priority boosted
|
|
// by this amount
|
|
int reliable_on_priority;
|
|
// results with a priority equal or greater than this value
|
|
// will be sent to reliable hosts
|
|
bool distinct_beta_apps;
|
|
// allow users to select beta apps independently
|
|
bool ended;
|
|
// Project has ended - tell clients to detach
|
|
int shmem_work_items;
|
|
// number of work items in shared memory
|
|
int feeder_query_size;
|
|
// number of work items to request in each feeder query
|
|
char httpd_user[256];
|
|
// user name under which web server runs (default: apache)
|
|
bool enable_assignment;
|
|
bool enable_vda;
|
|
double vda_host_timeout;
|
|
bool enable_assignment_multi;
|
|
bool job_size_matching;
|
|
bool dont_send_jobs;
|
|
|
|
//////////// STUFF RELEVANT ONLY TO SCHEDULER FOLLOWS ///////////
|
|
|
|
vector<regex_t> *ban_cpu;
|
|
vector<regex_t> *ban_os;
|
|
int daily_result_quota; // max results per day is this * mult
|
|
char debug_req_reply_dir[256];
|
|
// keep sched_request and sched_reply in files in this directory
|
|
double default_disk_max_used_gb;
|
|
double default_disk_max_used_pct;
|
|
double default_disk_min_free_gb;
|
|
vector<int> dont_search_host_for_userid;
|
|
bool dont_store_success_stderr;
|
|
int file_deletion_strategy;
|
|
// select method of automatically deleting files from host
|
|
int gpu_multiplier; // mult is NCPUS + this*NGPUS
|
|
bool ignore_delay_bound;
|
|
bool locality_scheduling;
|
|
double locality_scheduler_fraction;
|
|
bool locality_scheduling_sorted_order;
|
|
int locality_scheduling_wait_period;
|
|
int locality_scheduling_send_timeout;
|
|
vector<regex_t> *locality_scheduling_workunit_file;
|
|
vector<regex_t> *locality_scheduling_sticky_file;
|
|
bool matchmaker;
|
|
int max_download_urls_per_file;
|
|
int max_ncpus;
|
|
JOB_LIMITS max_jobs_in_progress;
|
|
int max_results_accepted;
|
|
// skip reported jobs beyond this limit
|
|
// (they'll get reported in the next RPC)
|
|
// This limits the memory usage of the scheduler;
|
|
// otherwise it can crash if the client is reporting thousands of jobs.
|
|
int max_wus_to_send; // max results per RPC is this * mult
|
|
int min_core_client_version;
|
|
int min_core_client_version_announced;
|
|
int min_core_client_upgrade_deadline;
|
|
int min_sendwork_interval;
|
|
int mm_min_slots;
|
|
int mm_max_slots;
|
|
double next_rpc_delay;
|
|
bool no_amd_k6;
|
|
// don't allow AMD K6 CPUs
|
|
bool no_vista_sandbox;
|
|
bool nowork_skip;
|
|
bool one_result_per_host_per_wu;
|
|
bool one_result_per_user_per_wu;
|
|
int reliable_max_avg_turnaround;
|
|
// max average turnaround for a host to be declared reliable
|
|
double reliable_max_error_rate;
|
|
// DEPRECATED
|
|
// max error rate for a host to be declared reliable
|
|
double reliable_reduced_delay_bound;
|
|
// Reduce the delay bounds for reliable hosts by this percent
|
|
char replace_download_url_by_timezone[256];
|
|
bool prefer_primary_platform;
|
|
// When selecting app versions,
|
|
// use the client's primary platform if a version exists.
|
|
// e.g. send 64-bit versions to 64-bit clients,
|
|
// rather than trying the 32-bit version to see if it's faster.
|
|
// Do this only if you're sure that the 64-bit versions are
|
|
// always faster than the corresponding 32-bit versions
|
|
double version_select_random_factor;
|
|
// in deciding what version is fastest,
|
|
// multiply projected FLOPS by a random var with mean 1 and this stddev.
|
|
int report_max;
|
|
bool request_time_stats_log;
|
|
bool resend_lost_results;
|
|
int sched_debug_level;
|
|
int scheduler_log_buffer;
|
|
char sched_lockfile_dir[256];
|
|
bool send_result_abort;
|
|
char symstore[256];
|
|
bool user_filter;
|
|
// send a job to a user only if wu.batch == user.id
|
|
// DEPRECATED: use assignment instead
|
|
bool workload_sim;
|
|
// Do workload simulation in deciding whether to send a result
|
|
bool estimate_flops_from_hav_pfc;
|
|
// Use host_app_version peak flop count rather than elapsed time
|
|
// to calculate projected_flops when choosing version.
|
|
|
|
// scheduler log flags
|
|
//
|
|
bool debug_array; // debug job-cache scheduling
|
|
bool debug_array_detail; // show slot-level info
|
|
bool debug_assignment;
|
|
bool debug_credit;
|
|
bool debug_edf_sim_detail; // show details of EDF sim
|
|
bool debug_edf_sim_workload; // show workload for EDF sim
|
|
bool debug_fcgi;
|
|
bool debug_client_files; // stuff related to sticky files on client
|
|
bool debug_handle_results;
|
|
bool debug_locality; // locality scheduling
|
|
bool debug_locality_lite; // locality scheduling Lite
|
|
bool debug_prefs;
|
|
bool debug_quota;
|
|
bool debug_request_details;
|
|
bool debug_request_headers;
|
|
bool debug_resend;
|
|
bool debug_send;
|
|
bool debug_user_messages;
|
|
bool debug_vda;
|
|
bool debug_version_select;
|
|
|
|
int parse(FILE*);
|
|
int parse_aux(FILE*);
|
|
int parse_file(const char *dir = 0);
|
|
|
|
int upload_path(const char*, char*);
|
|
int download_path(const char*, char*);
|
|
|
|
const char *project_path(const char *, ...);
|
|
};
|
|
|
|
extern SCHED_CONFIG config;
|
|
|
|
#endif
|