#include "config.h" #include #include #include #include "cluster.h" #include "ha.h" #include "xce/cluster_def.h" #include "xce/errnox.h" static int validate_host(cluster_t* c, const char* host, const char* key) { return 1; } static int handle_client_join(cluster_t* c, endpoint_t* ep, xce_request_t* request) { xce_reply_t reply; init_reply_by_req(&reply, request); reply.status = 0; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); return 0; } static int handle_worker_join(cluster_t* c, endpoint_t* ep, xce_request_t* request) { int idx; ep_private_t* priv = (ep_private_t*)ep->data; xce_reply_t reply; xce_peerinfo_t* from = request->from; int rc = 0; xce_joindata_t* joinmsg = NULL; TRACE("%s worker %s joining...", __func__, from->name); init_reply_by_req(&reply, request); if (!validate_host(c, from->name, from->key)) { ERROR("%s reject join of host %s", __func__, from->name); rc = -EPERM; reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); return rc; } joinmsg = request->body.joinmsg; lock(c->workermgr, RTTI_LOCK_WRITE); xce_run_hooks(&(c->hooksys), XCE_CONTROLLER_HOOK_WORKER, HOOK_BEFORE_EVENT, HOOK_DETAIL_WORKER_JOIN, c, from->name, 0); idx = add_worker(c, ep, from->name, joinmsg->data.attrs); reply.status = idx; reply.data_which = RequestReply_data_nodata; /* Look up the newly added worker so the persist hook can serialize it */ worker_t* w = NULL; if (idx >= 0) vget(c->workermgr, idx, (void**)&w); xce_run_hooks(&(c->hooksys), XCE_CONTROLLER_HOOK_WORKER, HOOK_AFTER_EVENT, HOOK_DETAIL_WORKER_JOIN, c, w); xce_reply(ep, c->myhost, &reply); unlock(c->workermgr); if (idx < 0) { ERROR("worker %s join failed, rc %d", from->name, idx); return idx; } priv->u.idx = idx; on_worker_join(c, ep); return 0; } static int handle_scheduler_join(cluster_t* c, endpoint_t* ep, xce_request_t* request) { xce_reply_t reply; ep_private_t* priv = (ep_private_t*)ep->data; int rc = 0; TRACE("%s scheduler join, ep %p", __func__, ep); init_reply_by_req(&reply, request); c->sched.ep = ep; priv->u.idx = 0; rc = on_scheduler_join(c); reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); return rc; } static int handle_controller_join(cluster_t* c, endpoint_t* ep, xce_request_t* request) { xce_reply_t reply; int rc = -ENOSYS; ERROR("we still have not support join of controller"); init_reply_by_req(&reply, request); reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); return rc; } static int handle_apiserver_join(cluster_t* c, endpoint_t* ep, xce_request_t* request) { xce_reply_t reply; xce_peerinfo_t* from = request->from; int rc = 0; TRACE("%s api server join, ep %p", __func__, ep); init_reply_by_req(&reply, request); rc = add_apiserver(c, from->name, ep); if (rc >= 0) { ep_private_t* priv = (ep_private_t*)ep->data; priv->u.idx = rc; on_apiserver_join(c, ep); } reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); if (rc < 0) { ERROR("api server %s join failed, rc %d", from->name, rc); return rc; } return 0; } static int handle_join_(cluster_t* c, endpoint_t* ep, xce_request_t* request) { ep_private_t* priv; int rc = -EINVAL; int role; xce_reply_t reply; xce_peerinfo_t* from = request->from; init_reply_by_req(&reply, request); priv = (ep_private_t*)ep->data; if (priv == NULL) { rc = -E_XCE_INTERNAL; ERROR("priv data of endpoint should not be NULL"); reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); return rc; } role = request->body.joinmsg->role; priv->role = role; TRACE("%s role %d join from %s", __func__, role, from->name); /* ── HA guard (P1-6) ──────────────────────────────────────── * A standby must never accept joins as if it were the primary. * If the primary address is known, tell the client where to go * (E_XCE_REDIRECT). If no primary is known yet, tell the client * to retry later (E_XCE_NOMASTER). */ if (ha_should_redirect(&c->ha)) { WARN("%s standby redirecting role-%d join from %s to %s:%u", __func__, role, from->name, c->ha.primary_host, c->ha.primary_port); rc = -E_XCE_REDIRECT; reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); return rc; } if (ha_no_primary_known(&c->ha)) { WARN("%s standby has no primary yet, rejecting role-%d join from %s", __func__, role, from->name); rc = -E_XCE_NOMASTER; reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); return rc; } switch (role) { case HostRole_client: rc = handle_client_join(c, ep, request); break; case HostRole_worker: rc = handle_worker_join(c, ep, request); break; case HostRole_scheduler: rc = handle_scheduler_join(c, ep, request); break; case HostRole_controller: rc = handle_controller_join(c, ep, request); break; case HostRole_apiserver: rc = handle_apiserver_join(c, ep, request); break; default: rc = -E_XCE_BAD_ROLE; ERROR("unknown role %d from host %s", role, from->name); reply.status = rc; reply.data_which = RequestReply_data_nodata; xce_reply(ep, c->myhost, &reply); break; } return rc; } int handle_join(cluster_t* c, endpoint_t* ep, xce_request_t* request) { int rc = 0; rc = handle_join_(c, ep, request); return rc; }