06c3dfc052
* Clean up the DVM so it continues to run even when applications error out and we would ordinarily abort the daemons. * Create a new errmgr component for the DVM to handle the differences. * Cleanup the DVM state component. * Add ORTE bindings directory and brief README * Pass a local tool index around to match jobs. * Pass the jobid on job completion. * Fix initialization logic. * Add framework for python wrapper. * Fix terminate-with-non-zero-exit behavior so it properly terminates only the indicated procs, notifies orte-submit, and orte-dvm continues executing. * Add some missing options to orte-dvm * Fix a bug in -host processing that caused us to ignore the #slots designator. Add a new attribute to indicate "do not expand the DVM" when submitting job spawn requests. * It actually makes no sense that we treat the termination of all children differently than terminating the children of a specific job - it only creates confusion over the difference in behavior. So terminate children the same way regardless. Extend the cmd_line utility to easily allow layering of command line definitions Catch up with ORTE interface change and make build more generic. Disable "fixed dvm" logic for now. Add another cmd_line function to merge a table of cmd line options with another one, reporting as errors any duplicate entries. Use this to allow orterun to reuse the orted_submit code Fix the "fixed_dvm" logic by ensuring we reset num_new_daemons to zero. Also ensure that the nidmap is sent with the first job so the downstream daemons get the node info. Remove a duplicate cmd line entry in orterun. Revise the DVM startup procedure to pass the nidmap only once, at the startup of the DVM. This reduces the overhead on each job launch and ensures that the nidmap doesn't get overwritten. Add new commands to get_orted_comm_cmd_str(). Move ORTE command line options to orte_globals.[ch]. Catch up with extra orte_submit_init parameter. Add example code. Add documentation. Bump version. The nidmap and routing data must be updated prior to propagating the xcast or else the xcast will fail. Fix the return code so it is something more expected when an error occurs. Ensure we get an error returned to us when we fail to launch for some reason. In this case, we will always get a launch_cb as we did indeed attempt to spawn it. The error code will be returned in the complete_cb. Fix the return code from orte_submit_job - it was returning the tracker index instead of "success". Take advantage of ORTE's pretty-print capabilities to provide a nice error output explaining why we failed to launch. Ensure we always get a launch_cb when we fail to launch, but no complete_cb as the job never launched. Extend the error reporting capability to job completion as well. Add index parameter to orte_submit_job(). Add orte_job_cancel and implement ORTE_DAEMON_TERMINATE_JOB_CMD. Factor out dvm termination. Parse the terminate option at tool level. Add error string for ORTE_ERR_JOB_CANCELLED. Add some safeguards. Cleanup and/of comments. Enable the return. Properly ORTE_DECLSPEC orte_submit_halt. Add orte_submit_halt and orte_submit_cancel to interface. Use the plm interface to terminate the job
456 строки
15 KiB
C
456 строки
15 KiB
C
/*
|
|
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
|
|
* University Research and Technology
|
|
* Corporation. All rights reserved.
|
|
* Copyright (c) 2004-2005 The University of Tennessee and The University
|
|
* of Tennessee Research Foundation. All rights
|
|
* reserved.
|
|
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
|
|
* University of Stuttgart. All rights reserved.
|
|
* Copyright (c) 2004-2005 The Regents of the University of California.
|
|
* All rights reserved.
|
|
* Copyright (c) 2010-2011 Cisco Systems, Inc. All rights reserved.
|
|
* Copyright (c) 2011-2013 Los Alamos National Security, LLC.
|
|
* All rights reserved.
|
|
* Copyright (c) 2014 Intel, Inc. All rights reserved.
|
|
* $COPYRIGHT$
|
|
*
|
|
* Additional copyrights may follow
|
|
*
|
|
* $HEADER$
|
|
*/
|
|
|
|
/** @file **/
|
|
|
|
#include "orte_config.h"
|
|
#include "orte/constants.h"
|
|
|
|
#include <stdio.h>
|
|
#ifdef HAVE_SYS_SIGNAL_H
|
|
#include <sys/signal.h>
|
|
#else
|
|
#include <signal.h>
|
|
#endif
|
|
|
|
#include "orte/mca/plm/plm_types.h"
|
|
#include "orte/util/error_strings.h"
|
|
#include "orte/runtime/orte_globals.h"
|
|
|
|
int orte_err2str(int errnum, const char **errmsg)
|
|
{
|
|
const char *retval;
|
|
switch (errnum) {
|
|
case ORTE_SUCCESS:
|
|
retval = "Success";
|
|
break;
|
|
case ORTE_ERR_RECV_LESS_THAN_POSTED:
|
|
retval = "Receive was less than posted size";
|
|
break;
|
|
case ORTE_ERR_RECV_MORE_THAN_POSTED:
|
|
retval = "Receive was greater than posted size";
|
|
break;
|
|
case ORTE_ERR_NO_MATCH_YET:
|
|
retval = "No match for receive posted";
|
|
break;
|
|
case ORTE_ERR_REQUEST:
|
|
retval = "Request error";
|
|
break;
|
|
case ORTE_ERR_NO_CONNECTION_ALLOWED:
|
|
retval = "No connection allowed";
|
|
break;
|
|
case ORTE_ERR_CONNECTION_REFUSED:
|
|
retval = "Connection refused";
|
|
break;
|
|
case ORTE_ERR_TYPE_MISMATCH:
|
|
retval = "Type mismatch";
|
|
break;
|
|
case ORTE_ERR_COMPARE_FAILURE:
|
|
retval = "Data comparison failure";
|
|
break;
|
|
case ORTE_ERR_COPY_FAILURE:
|
|
retval = "Data copy failure";
|
|
break;
|
|
case ORTE_ERR_PROC_STATE_MISSING:
|
|
retval = "The process state information is missing on the registry";
|
|
break;
|
|
case ORTE_ERR_PROC_EXIT_STATUS_MISSING:
|
|
retval = "The process exit status is missing on the registry";
|
|
break;
|
|
case ORTE_ERR_INDETERMINATE_STATE_INFO:
|
|
retval = "Request for state returned multiple responses";
|
|
break;
|
|
case ORTE_ERR_NODE_FULLY_USED:
|
|
retval = "All the slots on a given node have been used";
|
|
break;
|
|
case ORTE_ERR_INVALID_NUM_PROCS:
|
|
retval = "Multiple applications were specified, but at least one failed to specify the number of processes to run";
|
|
break;
|
|
case ORTE_ERR_SILENT:
|
|
if (orte_report_silent_errors) {
|
|
retval = "Silent error";
|
|
} else {
|
|
retval = NULL;
|
|
}
|
|
break;
|
|
case ORTE_ERR_ADDRESSEE_UNKNOWN:
|
|
retval = "A message is attempting to be sent to a process whose contact information is unknown";
|
|
break;
|
|
case ORTE_ERR_SYS_LIMITS_PIPES:
|
|
retval = "The system limit on number of pipes a process can open was reached";
|
|
break;
|
|
case ORTE_ERR_PIPE_SETUP_FAILURE:
|
|
retval = "A pipe could not be setup between a daemon and one of its local processes";
|
|
break;
|
|
case ORTE_ERR_SYS_LIMITS_CHILDREN:
|
|
retval = "The system limit on number of children a process can have was reached";
|
|
break;
|
|
case ORTE_ERR_FAILED_GET_TERM_ATTRS:
|
|
retval = "The I/O forwarding system was unable to get the attributes of your terminal";
|
|
break;
|
|
case ORTE_ERR_WDIR_NOT_FOUND:
|
|
retval = "The specified working directory could not be found";
|
|
break;
|
|
case ORTE_ERR_EXE_NOT_FOUND:
|
|
retval = "The specified executable could not be found";
|
|
break;
|
|
case ORTE_ERR_PIPE_READ_FAILURE:
|
|
retval = "A pipe could not be read";
|
|
break;
|
|
case ORTE_ERR_EXE_NOT_ACCESSIBLE:
|
|
retval = "The specified executable could not be executed";
|
|
break;
|
|
case ORTE_ERR_FAILED_TO_START:
|
|
retval = "The specified application failed to start";
|
|
break;
|
|
case ORTE_ERR_FILE_NOT_EXECUTABLE:
|
|
retval = "A system-required executable either could not be found or was not executable by this user";
|
|
break;
|
|
case ORTE_ERR_HNP_COULD_NOT_START:
|
|
retval = "Unable to start a daemon on the local node";
|
|
break;
|
|
case ORTE_ERR_SYS_LIMITS_SOCKETS:
|
|
retval = "The system limit on number of network connections a process can open was reached";
|
|
break;
|
|
case ORTE_ERR_SOCKET_NOT_AVAILABLE:
|
|
retval = "Unable to open a TCP socket for out-of-band communications";
|
|
break;
|
|
case ORTE_ERR_SYSTEM_WILL_BOOTSTRAP:
|
|
retval = "System will determine resources during bootstrap of daemons";
|
|
break;
|
|
case ORTE_ERR_RESTART_LIMIT_EXCEEDED:
|
|
retval = "Limit on number of process restarts was exceeded";
|
|
break;
|
|
case ORTE_ERR_INVALID_NODE_RANK:
|
|
retval = "Invalid node rank";
|
|
break;
|
|
case ORTE_ERR_INVALID_LOCAL_RANK:
|
|
retval = "Invalid local rank";
|
|
break;
|
|
case ORTE_ERR_UNRECOVERABLE:
|
|
retval = "Unrecoverable error";
|
|
break;
|
|
case ORTE_ERR_MEM_LIMIT_EXCEEDED:
|
|
retval = "Memory limit exceeded";
|
|
break;
|
|
case ORTE_ERR_HEARTBEAT_LOST:
|
|
retval = "Heartbeat lost";
|
|
break;
|
|
case ORTE_ERR_PROC_STALLED:
|
|
retval = "Proc appears to be stalled";
|
|
break;
|
|
case ORTE_ERR_NO_APP_SPECIFIED:
|
|
retval = "No application specified";
|
|
break;
|
|
case ORTE_ERR_NO_EXE_SPECIFIED:
|
|
retval = "No executable specified";
|
|
break;
|
|
case ORTE_ERR_COMM_DISABLED:
|
|
retval = "Communications have been disabled";
|
|
break;
|
|
case ORTE_ERR_FAILED_TO_MAP:
|
|
retval = "Unable to map job";
|
|
break;
|
|
case ORTE_ERR_TAKE_NEXT_OPTION:
|
|
if (orte_report_silent_errors) {
|
|
retval = "Next option";
|
|
} else {
|
|
retval = NULL;
|
|
}
|
|
break;
|
|
case ORTE_ERR_SENSOR_LIMIT_EXCEEDED:
|
|
retval = "Sensor limit exceeded";
|
|
break;
|
|
case ORTE_ERR_PROC_ENTRY_NOT_FOUND:
|
|
retval = "Proc entry not found";
|
|
break;
|
|
case ORTE_ERR_DATA_VALUE_NOT_FOUND:
|
|
retval = "Data not found";
|
|
break;
|
|
case ORTE_ERR_ALLOCATION_PENDING:
|
|
retval = "Allocation pending";
|
|
break;
|
|
case ORTE_ERR_NO_PATH_TO_TARGET:
|
|
retval = "No OOB path to target";
|
|
break;
|
|
case ORTE_ERR_OP_IN_PROGRESS:
|
|
retval = "Operation in progress";
|
|
break;
|
|
case ORTE_ERR_OPEN_CHANNEL_PEER_FAIL:
|
|
retval = "Open channel to peer failed";
|
|
break;
|
|
case ORTE_ERR_OPEN_CHANNEL_PEER_REJECT:
|
|
retval = "Open channel to peer was rejected";
|
|
break;
|
|
case ORTE_ERR_QOS_TYPE_UNSUPPORTED:
|
|
retval = "QoS type unsupported";
|
|
break;
|
|
case ORTE_ERR_QOS_ACK_WINDOW_FULL:
|
|
retval = "QoS ack window full";
|
|
break;
|
|
case ORTE_ERR_ACK_TIMEOUT_SENDER:
|
|
retval = "Send ack timed out";
|
|
break;
|
|
case ORTE_ERR_ACK_TIMEOUT_RECEIVER:
|
|
retval = "Recv ack timed out";
|
|
break;
|
|
case ORTE_ERR_LOST_MSG_IN_WINDOW:
|
|
retval = "Msg lost in window";
|
|
break;
|
|
case ORTE_ERR_CHANNEL_BUSY:
|
|
retval = "Channel busy";
|
|
break;
|
|
case ORTE_ERR_DUPLICATE_MSG:
|
|
retval = "Duplicate message";
|
|
break;
|
|
case ORTE_ERR_OUT_OF_ORDER_MSG:
|
|
retval = "Out of order message";
|
|
break;
|
|
case ORTE_ERR_OPEN_CHANNEL_DUPLICATE:
|
|
retval = "Duplicate channel open request";
|
|
break;
|
|
case ORTE_ERR_FORCE_SELECT:
|
|
retval = "Force select";
|
|
break;
|
|
case ORTE_ERR_JOB_CANCELLED:
|
|
retval = "Job cancelled";
|
|
break;
|
|
default:
|
|
if (orte_report_silent_errors) {
|
|
retval = "Unknown error";
|
|
} else {
|
|
retval = NULL;
|
|
}
|
|
}
|
|
|
|
*errmsg = retval;
|
|
return ORTE_SUCCESS;
|
|
}
|
|
|
|
const char *orte_job_state_to_str(orte_job_state_t state)
|
|
{
|
|
switch(state) {
|
|
case ORTE_JOB_STATE_UNDEF:
|
|
return "UNDEFINED";
|
|
case ORTE_JOB_STATE_INIT:
|
|
return "PENDING INIT";
|
|
case ORTE_JOB_STATE_INIT_COMPLETE:
|
|
return "INIT_COMPLETE";
|
|
case ORTE_JOB_STATE_ALLOCATE:
|
|
return "PENDING ALLOCATION";
|
|
case ORTE_JOB_STATE_ALLOCATION_COMPLETE:
|
|
return "ALLOCATION COMPLETE";
|
|
case ORTE_JOB_STATE_MAP:
|
|
return "PENDING MAPPING";
|
|
case ORTE_JOB_STATE_MAP_COMPLETE:
|
|
return "MAP COMPLETE";
|
|
case ORTE_JOB_STATE_SYSTEM_PREP:
|
|
return "PENDING FINAL SYSTEM PREP";
|
|
case ORTE_JOB_STATE_LAUNCH_DAEMONS:
|
|
return "PENDING DAEMON LAUNCH";
|
|
case ORTE_JOB_STATE_DAEMONS_LAUNCHED:
|
|
return "DAEMONS LAUNCHED";
|
|
case ORTE_JOB_STATE_DAEMONS_REPORTED:
|
|
return "ALL DAEMONS REPORTED";
|
|
case ORTE_JOB_STATE_VM_READY:
|
|
return "VM READY";
|
|
case ORTE_JOB_STATE_LAUNCH_APPS:
|
|
return "PENDING APP LAUNCH";
|
|
case ORTE_JOB_STATE_RUNNING:
|
|
return "RUNNING";
|
|
case ORTE_JOB_STATE_SUSPENDED:
|
|
return "SUSPENDED";
|
|
case ORTE_JOB_STATE_REGISTERED:
|
|
return "SYNC REGISTERED";
|
|
case ORTE_JOB_STATE_READY_FOR_DEBUGGERS:
|
|
return "READY FOR DEBUGGERS";
|
|
case ORTE_JOB_STATE_LOCAL_LAUNCH_COMPLETE:
|
|
return "LOCAL LAUNCH COMPLETE";
|
|
case ORTE_JOB_STATE_UNTERMINATED:
|
|
return "UNTERMINATED";
|
|
case ORTE_JOB_STATE_TERMINATED:
|
|
return "NORMALLY TERMINATED";
|
|
case ORTE_JOB_STATE_NOTIFY_COMPLETED:
|
|
return "NOTIFY COMPLETED";
|
|
case ORTE_JOB_STATE_NOTIFIED:
|
|
return "NOTIFIED";
|
|
case ORTE_JOB_STATE_ALL_JOBS_COMPLETE:
|
|
return "ALL JOBS COMPLETE";
|
|
case ORTE_JOB_STATE_ERROR:
|
|
return "ARTIFICIAL BOUNDARY - ERROR";
|
|
case ORTE_JOB_STATE_KILLED_BY_CMD:
|
|
return "KILLED BY INTERNAL COMMAND";
|
|
case ORTE_JOB_STATE_ABORTED:
|
|
return "ABORTED";
|
|
case ORTE_JOB_STATE_FAILED_TO_START:
|
|
return "FAILED TO START";
|
|
case ORTE_JOB_STATE_ABORTED_BY_SIG:
|
|
return "ABORTED BY SIGNAL";
|
|
case ORTE_JOB_STATE_ABORTED_WO_SYNC:
|
|
return "TERMINATED WITHOUT SYNC";
|
|
case ORTE_JOB_STATE_COMM_FAILED:
|
|
return "COMMUNICATION FAILURE";
|
|
case ORTE_JOB_STATE_SENSOR_BOUND_EXCEEDED:
|
|
return "SENSOR BOUND EXCEEDED";
|
|
case ORTE_JOB_STATE_CALLED_ABORT:
|
|
return "PROC CALLED ABORT";
|
|
case ORTE_JOB_STATE_HEARTBEAT_FAILED:
|
|
return "HEARTBEAT FAILED";
|
|
case ORTE_JOB_STATE_NEVER_LAUNCHED:
|
|
return "NEVER LAUNCHED";
|
|
case ORTE_JOB_STATE_ABORT_ORDERED:
|
|
return "ABORT IN PROGRESS";
|
|
case ORTE_JOB_STATE_NON_ZERO_TERM:
|
|
return "AT LEAST ONE PROCESS EXITED WITH NON-ZERO STATUS";
|
|
case ORTE_JOB_STATE_FAILED_TO_LAUNCH:
|
|
return "FAILED TO LAUNCH";
|
|
case ORTE_JOB_STATE_FORCED_EXIT:
|
|
return "FORCED EXIT";
|
|
case ORTE_JOB_STATE_DAEMONS_TERMINATED:
|
|
return "DAEMONS TERMINATED";
|
|
case ORTE_JOB_STATE_SILENT_ABORT:
|
|
return "ERROR REPORTED ELSEWHERE";
|
|
case ORTE_JOB_STATE_REPORT_PROGRESS:
|
|
return "REPORT PROGRESS";
|
|
case ORTE_JOB_STATE_ALLOC_FAILED:
|
|
return "ALLOCATION FAILED";
|
|
case ORTE_JOB_STATE_MAP_FAILED:
|
|
return "MAP FAILED";
|
|
case ORTE_JOB_STATE_CANNOT_LAUNCH:
|
|
return "CANNOT LAUNCH";
|
|
case ORTE_JOB_STATE_FT_CHECKPOINT:
|
|
return "FAULT TOLERANCE CHECKPOINT";
|
|
case ORTE_JOB_STATE_FT_CONTINUE:
|
|
return "FAULT TOLERANCE CONTINUE";
|
|
case ORTE_JOB_STATE_FT_RESTART:
|
|
return "FAULT TOLERANCE RESTART";
|
|
case ORTE_JOB_STATE_ANY:
|
|
return "ANY";
|
|
default:
|
|
return "UNKNOWN STATE!";
|
|
}
|
|
}
|
|
|
|
const char *orte_app_ctx_state_to_str(orte_app_state_t state)
|
|
{
|
|
switch(state) {
|
|
case ORTE_APP_STATE_UNDEF:
|
|
return "UNDEFINED";
|
|
case ORTE_APP_STATE_INIT:
|
|
return "PENDING INIT";
|
|
case ORTE_APP_STATE_ALL_MAPPED:
|
|
return "ALL MAPPED";
|
|
case ORTE_APP_STATE_RUNNING:
|
|
return "RUNNING";
|
|
case ORTE_APP_STATE_COMPLETED:
|
|
return "COMPLETED";
|
|
default:
|
|
return "UNKNOWN STATE!";
|
|
}
|
|
}
|
|
|
|
const char *orte_proc_state_to_str(orte_proc_state_t state)
|
|
{
|
|
switch(state) {
|
|
case ORTE_PROC_STATE_UNDEF:
|
|
return "UNDEFINED";
|
|
case ORTE_PROC_STATE_INIT:
|
|
return "INITIALIZED";
|
|
case ORTE_PROC_STATE_RESTART:
|
|
return "RESTARTING";
|
|
case ORTE_PROC_STATE_TERMINATE:
|
|
return "MARKED FOR TERMINATION";
|
|
case ORTE_PROC_STATE_RUNNING:
|
|
return "RUNNING";
|
|
case ORTE_PROC_STATE_REGISTERED:
|
|
return "SYNC REGISTERED";
|
|
case ORTE_PROC_STATE_IOF_COMPLETE:
|
|
return "IOF COMPLETE";
|
|
case ORTE_PROC_STATE_WAITPID_FIRED:
|
|
return "WAITPID FIRED";
|
|
case ORTE_PROC_STATE_UNTERMINATED:
|
|
return "UNTERMINATED";
|
|
case ORTE_PROC_STATE_TERMINATED:
|
|
return "NORMALLY TERMINATED";
|
|
case ORTE_PROC_STATE_ERROR:
|
|
return "ARTIFICIAL BOUNDARY - ERROR";
|
|
case ORTE_PROC_STATE_KILLED_BY_CMD:
|
|
return "KILLED BY INTERNAL COMMAND";
|
|
case ORTE_PROC_STATE_ABORTED:
|
|
return "ABORTED";
|
|
case ORTE_PROC_STATE_FAILED_TO_START:
|
|
return "FAILED TO START";
|
|
case ORTE_PROC_STATE_ABORTED_BY_SIG:
|
|
return "ABORTED BY SIGNAL";
|
|
case ORTE_PROC_STATE_TERM_WO_SYNC:
|
|
return "TERMINATED WITHOUT SYNC";
|
|
case ORTE_PROC_STATE_COMM_FAILED:
|
|
return "COMMUNICATION FAILURE";
|
|
case ORTE_PROC_STATE_SENSOR_BOUND_EXCEEDED:
|
|
return "SENSOR BOUND EXCEEDED";
|
|
case ORTE_PROC_STATE_CALLED_ABORT:
|
|
return "CALLED ABORT";
|
|
case ORTE_PROC_STATE_HEARTBEAT_FAILED:
|
|
return "HEARTBEAT FAILED";
|
|
case ORTE_PROC_STATE_MIGRATING:
|
|
return "MIGRATING";
|
|
case ORTE_PROC_STATE_CANNOT_RESTART:
|
|
return "CANNOT BE RESTARTED";
|
|
case ORTE_PROC_STATE_TERM_NON_ZERO:
|
|
return "EXITED WITH NON-ZERO STATUS";
|
|
case ORTE_PROC_STATE_FAILED_TO_LAUNCH:
|
|
return "FAILED TO LAUNCH";
|
|
case ORTE_PROC_STATE_UNABLE_TO_SEND_MSG:
|
|
return "UNABLE TO SEND MSG";
|
|
case ORTE_PROC_STATE_LIFELINE_LOST:
|
|
return "LIFELINE LOST";
|
|
case ORTE_PROC_STATE_ANY:
|
|
return "ANY";
|
|
default:
|
|
return "UNKNOWN STATE!";
|
|
}
|
|
}
|
|
|
|
const char *orte_node_state_to_str(orte_node_state_t state)
|
|
{
|
|
switch(state) {
|
|
case ORTE_NODE_STATE_UNDEF:
|
|
return "UNDEF";
|
|
case ORTE_NODE_STATE_UNKNOWN:
|
|
return "UNKNOWN";
|
|
case ORTE_NODE_STATE_DOWN:
|
|
return "DOWN";
|
|
case ORTE_NODE_STATE_UP:
|
|
return "UP";
|
|
case ORTE_NODE_STATE_REBOOT:
|
|
return "REBOOT";
|
|
case ORTE_NODE_STATE_DO_NOT_USE:
|
|
return "DO_NOT_USE";
|
|
case ORTE_NODE_STATE_NOT_INCLUDED:
|
|
return "NOT_INCLUDED";
|
|
case ORTE_NODE_STATE_ADDED:
|
|
return "ADDED";
|
|
default:
|
|
return "UNKNOWN STATE!";
|
|
}
|
|
}
|