myxce/src/include/xce/cluster_def.h

#ifndef XCE_CLUSTER_DEF_H
#define XCE_CLUSTER_DEF_H

#include "config.h"

#include <pthread.h>
#include <rtti.h>
#include <stdint.h>

#include "delay.h"
#include "hook.h"
#include "htab.h"
#include "kvmgr.h"
#include "myconf.h"
#include "mylock.h"
#include "netio.h"
#include "persist.h"
#include "vtaskmgr.h"

/* ha.h forward-declares cluster_t; including it here is safe (no circular dep) */
#include "ha.h"
#include "xce/types.h"
#include "xlist.h"
#include "zarray.h"

#define N_CONTROLLER_MAX (16)
#define N_APISERVER_MAX  (16)

enum {
    XCE_CONTROLLER_HOOK_JOB = 0,
    XCE_CONTROLLER_HOOK_WORKER,
    XCE_CONTROLLER_HOOK_SCHEDULER,
    XCE_CONTROLLER_HOOK_CLUSTER,
    XCE_CONTROLLER_HOOK_RESOURCE,
    XCE_CONTROLLER_HOOK_QUEUE,
    XCE_CONTROLLER_HOOK_USERGROUP,
    XCE_CONTROLLER_HOOK_HOSTGROUP,
    XCE_CONTROLLER_HOOK_MAX /* keep this as the max */
};

enum {
    HOOK_DETAIL_JOB_NEW = 0,
    HOOK_DETAIL_JOB_START,
    HOOK_DETAIL_JOB_EXIT,
    HOOK_DETAIL_JOB_FAIL,
    HOOK_DETAIL_JOB_UPDATE,
    HOOK_DETAIL_JOB_CKPT,
    HOOK_DETAIL_SETPREASON,
    HOOK_DETAIL_CLRPREASON,
    HOOK_DETAIL_WORKER_JOIN,
    HOOK_DETAIL_WORKER_DEAD,
    HOOK_DETAIL_WORKER_UPDATE,
    HOOK_DETAIL_SCHEDULER_JOIN,
    HOOK_DETAIL_SCHEDULER_DEAD,
    HOOK_DETAIL_CLUSTER_INIT,
    HOOK_DETAIL_CLUSTER_SHUTDOWN,
    HOOK_DETAIL_RESOURCE_UPDATE,
    HOOK_DETAIL_RESOURCE_LOC_UPDATE,
    HOOK_DETAIL_QUEUE_UPDATE,
    HOOK_DETAIL_USERGROUP_UPDATE,
    HOOK_DETAIL_HOSTGROUP_UPDATE,
    HOOK_DETAIL_JOB_RUSAGE_UPDATE,
    HOOK_DETAIL_JOB_STOP,
    HOOK_DETAIL_JOB_RESUME
};

#define XCE_NS_CONFIG (0)
#define XCE_NS_UGROUP (1)
#define XCE_NS_HGROUP (2)
#define XCE_NS_MAX    (2)

typedef struct {
    int role;
    union {
        int idx;
    } u;
} ep_private_t;

typedef struct {
    endpoint_t* ep;
} controller_t;

typedef struct {
} metrics_t;

typedef struct {
    mylock_t    lock;
    endpoint_t* ep;
} scheduler_t;

typedef struct {
    mylock_t    lock;
    char*       name;
    endpoint_t* ep;
} cluster_apiserver_t;

typedef struct cluster_s {
    myconf_t*           myconf;
    xce_peerinfo_t*     myhost;       /* points at myhost_info once initialised */
    xce_peerinfo_t      myhost_info;  /* storage; name set from myconf->name */
    hooksys_t           hooksys;
    iomgr_t             iomgr;
    zarray_t            endpoints;
    endpoint_t          ep;
    zarray_t            configs;
    zarray_t            sharedinst;
    controller_t        controllers[N_CONTROLLER_MAX];
    cluster_apiserver_t apiservers[N_APISERVER_MAX];
    int                 myidx;
    int                 master;
    int                 ncontrollers;
    delaymgr_t          delaymgr;
    scheduler_t         sched;
    var                 jobmgr;
    var                 resmgr;
    var                 workermgr;
    var                 qmgr;
    var                 configmgr;
    vtaskmgr_t          vtaskmgr;
    cbuf_t              wbuf;
    int                 running;
    persist_t           persist;   /* Persistence module — P0-F */
    ha_state_t          ha;        /* HA role management — P1-5 */
    _Atomic int         restart_requested;           /* set by OP_upgradeConfig handler */
    pthread_t           heartbeat_listener_tid;      /* pre-Phase-2 SSL listener thread */
    _Atomic int         heartbeat_listener_running;
    _Atomic int         lost_and_found_job_count;    /* incremented during persist_recover() */

    /* Performance-tuning parameters — initialized by init_controller_tunables()
     * in cluster_init_phase1 so they are available before cluster_init_phase2.
     * TODO: wire to cluster_config once config generation tool supports @13-@17. */
    struct {
        int jobmgr_nhash;     /* hash slots for job table */
        int jobmgr_delta;     /* growth step for job table */
        int workermgr_nhash;  /* hash slots for worker table */
        int workermgr_delta;  /* growth step for worker table */
        int qmgr_nhash;       /* hash slots for queue table */
        int qmgr_delta;       /* growth step for queue table */
        int resmgr_nhash;     /* hash slots for resource table */
        int resmgr_delta;     /* growth step for resource table */
        int resmgr_inst;      /* growth step for shared-instance array */
        int xstrings_nhash;   /* hash slots for xstring table */
    } tunables;
} cluster_t;

#endif /* XCE_CLUSTER_DEF_H */