forked from lei.chen/myxce
145 lines
6.9 KiB
C
145 lines
6.9 KiB
C
#if !defined(_CLUSTER_H_)
|
|
|
|
#define _CLUSTER_H_ 1
|
|
#include "config.h"
|
|
|
|
#include <pthread.h>
|
|
#include <stdbool.h>
|
|
#include <stdint.h>
|
|
|
|
#include "hook.h"
|
|
#include "htab.h"
|
|
#include "kvmgr.h"
|
|
#include "myconf.h"
|
|
#include "mylock.h"
|
|
#include "netio.h"
|
|
#include "vtaskmgr.h"
|
|
#include "worker.h"
|
|
#include "xce/cluster_def.h"
|
|
#include "xce/types.h"
|
|
#include "xlist.h"
|
|
#include "zarray.h"
|
|
|
|
typedef int (*collect_check_fn_t)(void*);
|
|
|
|
extern int check_controller(void*);
|
|
extern int cluster_init_phase1(cluster_t*);
|
|
extern int cluster_open_listener(cluster_t*);
|
|
extern int cluster_register_listener(cluster_t*);
|
|
extern int cluster_init_phase2(cluster_t*);
|
|
extern int cluster_init(cluster_t*);
|
|
extern int cluster_start(cluster_t*);
|
|
extern int cluster_run(cluster_t*);
|
|
extern void cluster_fini(cluster_t*);
|
|
extern int register_types(cluster_t*);
|
|
extern int worker_ctrl(cluster_t*, xce_hostctrl_t*);
|
|
extern xce_resource_t* get_resource(cluster_t*, char*);
|
|
extern int add_to_xlist(xlist_t*, void*);
|
|
extern int add_worker_to_xlist(xlist_t*, void*);
|
|
extern int add_queue_to_xlist(xlist_t*, void*);
|
|
extern int add_job_to_xlist(xlist_t*, void*);
|
|
extern void resource_killer(void*);
|
|
extern int add_resource(void*, char*, char*, int, xce_order_t, int, int);
|
|
|
|
extern int update_resvals(cluster_t*, worker_t*, int, xce_resval_t*);
|
|
|
|
extern int load_jobs(void*);
|
|
extern int restore_job(cluster_t*, uint64_t, const void*, size_t, uint64_t);
|
|
extern int update_job(void*, uint64_t, xce_kv_list_t*);
|
|
extern int do_job_started(void*, uint64_t, int, int*, int, int*);
|
|
|
|
extern queue_t* create_queue(cluster_t*, xce_queue_t*);
|
|
extern int build_queues(void*);
|
|
extern void queue_fini(queue_t*);
|
|
extern int queue_ctrl(cluster_t*, xce_queuectrl_t*);
|
|
|
|
extern int xce_update_data(endpoint_t*, cluster_t*, int, int, int*, int,
|
|
xce_kv_t*);
|
|
extern int on_scheduler_join(cluster_t*);
|
|
extern int on_apiserver_join(cluster_t*, endpoint_t*);
|
|
|
|
extern int xce_push_reslist(endpoint_t*, cluster_t*, int, xce_resource_t**);
|
|
extern int xce_push_confvars(endpoint_t*, cluster_t*, int, xce_config_t**);
|
|
extern int xce_push_workerlist(endpoint_t*, cluster_t*, int, xce_worker_t**);
|
|
extern int xce_push_joblist(endpoint_t*, cluster_t*, int, xce_job_t**);
|
|
extern int xce_push_queuelist(endpoint_t*, cluster_t*, int, xce_queue_t**);
|
|
extern int xcesched_push_job(cluster_t*, uint64_t);
|
|
extern int on_worker_join(void*, endpoint_t*);
|
|
extern int get_resval(cluster_t*, char*, char*, xce_val_t*);
|
|
extern int handle_push(struct capn*, cluster_t*, endpoint_t*, struct Request*);
|
|
extern int do_get_op(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int do_update_op(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int handle_join(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int handle_reply(endpoint_t*, struct Request*, cluster_t*);
|
|
extern int handle_jobdispatch(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int handle_jobsubmit(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int handle_worker_leave(cluster_t*, int);
|
|
extern int mark_worker_unavail(cluster_t*, worker_t*);
|
|
extern int mark_worker_back(cluster_t*, worker_t*);
|
|
extern int handle_scheduler_leave(cluster_t*);
|
|
extern int handle_clusterconf(cluster_t*, endpoint_t*, xce_request_t*);
|
|
|
|
extern void collect_confvars(cluster_t*, collect_check_fn_t, xlist_t*);
|
|
extern void collect_queues(cluster_t*, collect_check_fn_t, xlist_t*);
|
|
extern void collect_resources(cluster_t*, collect_check_fn_t, xlist_t*);
|
|
extern void collect_jobs(cluster_t*, collect_check_fn_t, xlist_t*);
|
|
extern void collect_workers(cluster_t*, collect_check_fn_t, xlist_t*);
|
|
|
|
/* ha-convergence-recovery (3.1-3.3): commit-watermark egress gate for
|
|
* scheduler-bound job exports. Defined in dataex.c.
|
|
* job_submit_committed — true iff the job's submit entry is committed
|
|
* (submit_lsn == 0 || submit_lsn <= committed_lsn);
|
|
* one lock-free atomic load + compare.
|
|
* collect_jobs_committed — identical contract to collect_jobs, additionally
|
|
* excluding jobs whose submit entry is uncommitted.
|
|
* For scheduler-bound callers ONLY; status/API read
|
|
* paths keep plain collect_jobs (Req 3.4). */
|
|
extern bool job_submit_committed(cluster_t*, job_t*);
|
|
extern void collect_jobs_committed(cluster_t*, collect_check_fn_t, xlist_t*);
|
|
extern int get_queue_exec(cluster_t*, char*, char**, char**);
|
|
extern void unlock_collect_queues(xlist_t*);
|
|
extern void unlock_collect_jobs(xlist_t*);
|
|
extern void unlock_collect_workers(xlist_t*);
|
|
|
|
extern int on_job_start(cluster_t*, job_pidinfo_t*);
|
|
extern int on_job_finish(cluster_t*, xce_jobfinish_t*);
|
|
extern int on_job_fail(cluster_t*, job_failinfo_t*);
|
|
extern int on_job_stop(cluster_t*, uint64_t);
|
|
extern int on_job_resume(cluster_t*, uint64_t);
|
|
extern int on_job_postdone(cluster_t*, endpoint_t*, uint64_t);
|
|
extern int on_job_ckpt(cluster_t*, endpoint_t*, xce_ckptinfo_t*);
|
|
extern int notify(cluster_t*, endpoint_t*, xce_notify_data_t*);
|
|
extern int check_worker_(xce_worker_t*);
|
|
extern int update_resval_list(cluster_t*, worker_t*, xce_resval_list_t*);
|
|
extern int do_jobcontrol_op(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int job_ctrl(cluster_t*, xce_jobctrl_t*);
|
|
extern int update_job_rusages(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int update_host_runusage(xce_job_runusage_t**, char*, int,
|
|
xce_rusage_t**);
|
|
extern int update_pids(job_pidinfo_t**, int, int*);
|
|
extern int update_pgids(job_pidinfo_t**, int, int*);
|
|
extern int get_ep_role(endpoint_t*);
|
|
extern worker_t* get_ep_worker(cluster_t*, endpoint_t*);
|
|
|
|
extern int check_worker_(xce_worker_t*);
|
|
extern int update_resval_list(cluster_t*, worker_t*, xce_resval_list_t*);
|
|
extern int do_jobcontrol_op(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int job_ctrl(cluster_t*, xce_jobctrl_t*);
|
|
extern int do_queuecontrol_op(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int do_hostcontrol_op(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int do_runtasks(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int on_task_start(cluster_t*, xce_task_pidinfo_t*);
|
|
extern int on_task_fail(cluster_t*, xce_task_failinfo_t*);
|
|
extern int on_task_finish(cluster_t*, xce_task_finish_t*);
|
|
|
|
extern int xce_controller_res_init(cluster_t*);
|
|
extern int map_resource_to_host(char*, cluster_t*);
|
|
extern int release_resource_from_worker(char*, cluster_t*);
|
|
extern int collect_resval(var, zarray_t*, xlist_t*);
|
|
extern int do_set_preason(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int do_clr_preason(cluster_t*, endpoint_t*, xce_request_t*);
|
|
extern int handle_apiserver_leave(cluster_t*, endpoint_t*);
|
|
extern int add_apiserver(cluster_t*, char*, endpoint_t*);
|
|
extern int apiserver_init(cluster_t*);
|
|
|
|
#endif
|