a200e4f865
*** THIS RFC INCLUDES A MINOR CHANGE TO THE MPI-RTE INTERFACE *** Note: during the course of this work, it was necessary to completely separate the MPI and RTE progress engines. There were multiple places in the MPI layer where ORTE_WAIT_FOR_COMPLETION was being used. A new OMPI_WAIT_FOR_COMPLETION macro was created (defined in ompi/mca/rte/rte.h) that simply cycles across opal_progress until the provided flag becomes false. Places where the MPI layer blocked waiting for RTE to complete an event have been modified to use this macro. *************************************************************************************** I am reissuing this RFC because of the time that has passed since its original release. Since its initial release and review, I have debugged it further to ensure it fully supports tests like loop_spawn. It therefore seems ready for merge back to the trunk. Given its prior review, I have set the timeout for one week. The code is in https://bitbucket.org/rhc/ompi-oob2 WHAT: Rewrite of ORTE OOB WHY: Support asynchronous progress and a host of other features WHEN: Wed, August 21 SYNOPSIS: The current OOB has served us well, but a number of limitations have been identified over the years. Specifically: * it is only progressed when called via opal_progress, which can lead to hangs or recursive calls into libevent (which is not supported by that code) * we've had issues when multiple NICs are available as the code doesn't "shift" messages between transports - thus, all nodes had to be available via the same TCP interface. * the OOB "unloads" incoming opal_buffer_t objects during the transmission, thus preventing use of OBJ_RETAIN in the code when repeatedly sending the same message to multiple recipients * there is no failover mechanism across NICs - if the selected NIC (or its attached switch) fails, we are forced to abort * only one transport (i.e., component) can be "active" The revised OOB resolves these problems: * async progress is used for all application processes, with the progress thread blocking in the event library * each available TCP NIC is supported by its own TCP module. The ability to asynchronously progress each module independently is provided, but not enabled by default (a runtime MCA parameter turns it "on") * multi-address TCP NICs (e.g., a NIC with both an IPv4 and IPv6 address, or with virtual interfaces) are supported - reachability is determined by comparing the contact info for a peer against all addresses within the range covered by the address/mask pairs for the NIC. * a message that arrives on one TCP NIC is automatically shifted to whatever NIC that is connected to the next "hop" if that peer cannot be reached by the incoming NIC. If no TCP module will reach the peer, then the OOB attempts to send the message via all other available components - if none can reach the peer, then an "error" is reported back to the RML, which then calls the errmgr for instructions. * opal_buffer_t now conforms to standard object rules re OBJ_RETAIN as we no longer "unload" the incoming object * NIC failure is reported to the TCP component, which then tries to resend the message across any other available TCP NIC. If that doesn't work, then the message is given back to the OOB base to try using other components. If all that fails, then the error is reported to the RML, which reports to the errmgr for instructions * obviously from the above, multiple OOB components (e.g., TCP and UD) can be active in parallel * the matching code has been moved to the RML (and out of the OOB/TCP component) so it is independent of transport * routing is done by the individual OOB modules (as opposed to the RML). Thus, both routed and non-routed transports can simultaneously be active * all blocking send/recv APIs have been removed. Everything operates asynchronously. KNOWN LIMITATIONS: * although provision is made for component failover as described above, the code for doing so has not been fully implemented yet. At the moment, if all connections for a given peer fail, the errmgr is notified of a "lost connection", which by default results in termination of the job if it was a lifeline * the IPv6 code is present and compiles, but is not complete. Since the current IPv6 support in the OOB doesn't work anyway, I don't consider this a blocker * routing is performed at the individual module level, yet the active routed component is selected on a global basis. We probably should update that to reflect that different transports may need/choose to route in different ways * obviously, not every error path has been tested nor necessarily covered * determining abnormal termination is more challenging than in the old code as we now potentially have multiple ways of connecting to a process. Ideally, we would declare "connection failed" when *all* transports can no longer reach the process, but that requires some additional (possibly complex) code. For now, the code replicates the old behavior only somewhat modified - i.e., if a module sees its connection fail, it checks to see if it is a lifeline. If so, it notifies the errmgr that the lifeline is lost - otherwise, it notifies the errmgr that a non-lifeline connection was lost. * reachability is determined solely on the basis of a shared subnet address/mask - more sophisticated algorithms (e.g., the one used in the tcp btl) are required to handle routing via gateways * the RML needs to assign sequence numbers to each message on a per-peer basis. The receiving RML will then deliver messages in order, thus preventing out-of-order messaging in the case where messages travel across different transports or a message needs to be redirected/resent due to failure of a NIC This commit was SVN r29058.
407 строки
14 KiB
C
407 строки
14 KiB
C
/*
|
|
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
|
|
* University Research and Technology
|
|
* Corporation. All rights reserved.
|
|
* Copyright (c) 2004-2011 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) 2007 Cisco Systems, Inc. All rights reserved.
|
|
* Copyright (c) 2011-2012 Los Alamos National Security, LLC. All rights
|
|
* reserved.
|
|
* $COPYRIGHT$
|
|
*
|
|
* Additional copyrights may follow
|
|
*
|
|
* $HEADER$
|
|
*/
|
|
|
|
#include "orte_config.h"
|
|
#include "orte/constants.h"
|
|
#include "orte/types.h"
|
|
|
|
#include "opal/dss/dss.h"
|
|
|
|
#include "orte/mca/errmgr/errmgr.h"
|
|
#include "orte/mca/ess/ess.h"
|
|
#include "orte/mca/odls/odls_types.h"
|
|
#include "orte/mca/rml/rml.h"
|
|
#include "orte/runtime/orte_globals.h"
|
|
#include "orte/runtime/orte_wait.h"
|
|
|
|
#include "orte/mca/routed/base/base.h"
|
|
|
|
void orte_routed_base_xcast_routing(orte_grpcomm_collective_t *coll,
|
|
opal_list_t *my_children)
|
|
{
|
|
opal_list_item_t *item;
|
|
orte_routed_tree_t *child;
|
|
orte_namelist_t *nm;
|
|
int i;
|
|
orte_proc_t *proc;
|
|
orte_job_t *daemons;
|
|
|
|
/* if we are the HNP and an abnormal termination is underway,
|
|
* then send it directly to everyone
|
|
*/
|
|
if (ORTE_PROC_IS_HNP) {
|
|
if (orte_abnormal_term_ordered) {
|
|
daemons = orte_get_job_data_object(ORTE_PROC_MY_NAME->jobid);
|
|
for (i=1; i < daemons->procs->size; i++) {
|
|
if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(daemons->procs, i))) {
|
|
continue;
|
|
}
|
|
/* exclude anyone known not alive */
|
|
if (proc->alive) {
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = ORTE_PROC_MY_NAME->jobid;
|
|
nm->name.vpid = proc->name.vpid;
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
}
|
|
}
|
|
} else {
|
|
/* the xcast always goes to our children */
|
|
for (item = opal_list_get_first(my_children);
|
|
item != opal_list_get_end(my_children);
|
|
item = opal_list_get_next(item)) {
|
|
child = (orte_routed_tree_t*)item;
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = ORTE_PROC_MY_NAME->jobid;
|
|
nm->name.vpid = child->vpid;
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
}
|
|
}
|
|
} else {
|
|
/* I am a daemon - route to my children */
|
|
for (item = opal_list_get_first(my_children);
|
|
item != opal_list_get_end(my_children);
|
|
item = opal_list_get_next(item)) {
|
|
child = (orte_routed_tree_t*)item;
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = ORTE_PROC_MY_NAME->jobid;
|
|
nm->name.vpid = child->vpid;
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
}
|
|
}
|
|
}
|
|
|
|
void orte_routed_base_coll_relay_routing(orte_grpcomm_collective_t *coll)
|
|
{
|
|
opal_list_item_t *item, *itm;
|
|
orte_namelist_t *nm, *n2, *n3;
|
|
bool dup;
|
|
orte_job_t *jdata;
|
|
orte_proc_t *proc;
|
|
|
|
if (ORTE_PROC_IS_HNP) {
|
|
/* nobody to send to */
|
|
return;
|
|
}
|
|
/* if we are a daemon, then we look at the list of
|
|
* participants. If there is a wildcard, then we
|
|
* know that all procs are participating, so we
|
|
* can send it to our parent. If not, then we have
|
|
* to send the collective to the daemon hosting
|
|
* the participating proc
|
|
*/
|
|
for (item = opal_list_get_first(&coll->participants);
|
|
item != opal_list_get_end(&coll->participants);
|
|
item = opal_list_get_next(item)) {
|
|
n2 = (orte_namelist_t*)item;
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = ORTE_PROC_MY_NAME->jobid;
|
|
dup = false;
|
|
if (ORTE_VPID_WILDCARD == n2->name.vpid) {
|
|
nm->name.vpid = ORTE_PROC_MY_PARENT->vpid;
|
|
} else {
|
|
jdata = orte_get_job_data_object(n2->name.jobid);
|
|
proc = (orte_proc_t*)opal_pointer_array_get_item(jdata->procs, n2->name.vpid);
|
|
if (NULL == proc || NULL == proc->node || NULL == proc->node->daemon) {
|
|
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
|
|
continue;
|
|
}
|
|
nm->name.vpid = proc->node->daemon->name.vpid;
|
|
}
|
|
/* if it is me, then ignore */
|
|
if (nm->name.vpid == ORTE_PROC_MY_NAME->vpid) {
|
|
dup = true;
|
|
} else {
|
|
/* if it is already on the list, we ignore */
|
|
for (itm = opal_list_get_first(&coll->targets);
|
|
itm != opal_list_get_end(&coll->targets);
|
|
itm = opal_list_get_next(itm)) {
|
|
n3 = (orte_namelist_t*)itm;
|
|
if (n3->name.vpid == nm->name.vpid) {
|
|
/* duplicate */
|
|
dup = true;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
if (dup) {
|
|
OBJ_RELEASE(nm);
|
|
} else {
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
}
|
|
}
|
|
}
|
|
|
|
void orte_routed_base_coll_complete_routing(orte_grpcomm_collective_t *coll)
|
|
{
|
|
opal_list_item_t *item;
|
|
orte_namelist_t *nm, *n2;
|
|
int i;
|
|
orte_proc_t *proc;
|
|
|
|
if (ORTE_PROC_IS_HNP) {
|
|
/* send it to everyone that participated */
|
|
for (item = opal_list_get_first(&coll->participants);
|
|
item != opal_list_get_end(&coll->participants);
|
|
item = opal_list_get_next(item)) {
|
|
n2 = (orte_namelist_t*)item;
|
|
/* if the vpid is wildcard, then the result will go
|
|
* to everyone in the job via xcast, so just carry it
|
|
* across
|
|
*/
|
|
if (ORTE_VPID_WILDCARD == n2->name.vpid) {
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = n2->name.jobid;
|
|
nm->name.vpid = n2->name.vpid;
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
} else {
|
|
/* only include it if the proc is local to us */
|
|
for (i=0; i < orte_local_children->size; i++) {
|
|
if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(orte_local_children, i))) {
|
|
continue;
|
|
}
|
|
if (proc->name.jobid == n2->name.jobid &&
|
|
proc->name.vpid == n2->name.vpid) {
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = n2->name.jobid;
|
|
nm->name.vpid = n2->name.vpid;
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
} else {
|
|
/* if the participants are wildcard, then the HNP will
|
|
* be sending the result to the procs via xcast. For all
|
|
* other cases, handle our own local children
|
|
*/
|
|
for (item = opal_list_get_first(&coll->participants);
|
|
item != opal_list_get_end(&coll->participants);
|
|
item = opal_list_get_next(item)) {
|
|
n2 = (orte_namelist_t*)item;
|
|
if (ORTE_VPID_WILDCARD == n2->name.vpid) {
|
|
continue;
|
|
}
|
|
for (i=0; i < orte_local_children->size; i++) {
|
|
if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(orte_local_children, i))) {
|
|
continue;
|
|
}
|
|
if (proc->name.jobid == n2->name.jobid &&
|
|
proc->name.vpid == n2->name.vpid) {
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = n2->name.jobid;
|
|
nm->name.vpid = n2->name.vpid;
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
void orte_routed_base_coll_peers(orte_grpcomm_collective_t *coll,
|
|
opal_list_t *my_children)
|
|
{
|
|
opal_list_item_t *item;
|
|
orte_routed_tree_t *child;
|
|
orte_namelist_t *nm;
|
|
|
|
/* tree-based systems require input from their children */
|
|
for (item = opal_list_get_first(my_children);
|
|
item != opal_list_get_end(my_children);
|
|
item = opal_list_get_next(item)) {
|
|
child = (orte_routed_tree_t*)item;
|
|
nm = OBJ_NEW(orte_namelist_t);
|
|
nm->name.jobid = ORTE_PROC_MY_NAME->jobid;
|
|
nm->name.vpid = child->vpid;
|
|
opal_list_append(&coll->targets, &nm->super);
|
|
}
|
|
}
|
|
|
|
|
|
static void report_sync(int status, orte_process_name_t* sender,
|
|
opal_buffer_t *buffer,
|
|
orte_rml_tag_t tag, void *cbdata)
|
|
{
|
|
bool *sync_waiting = (bool*)cbdata;
|
|
|
|
/* just copy the payload to the sync_buf */
|
|
opal_dss.copy_payload(orte_process_info.sync_buf, buffer);
|
|
/* flag as complete */
|
|
*sync_waiting = false;
|
|
}
|
|
|
|
int orte_routed_base_register_sync(bool setup)
|
|
{
|
|
opal_buffer_t *buffer;
|
|
int rc;
|
|
orte_daemon_cmd_flag_t command;
|
|
char *rml_uri;
|
|
uint8_t flag;
|
|
bool sync_waiting;
|
|
|
|
if (orte_abnormal_term_ordered) {
|
|
/* if we are abnormally terminating, don't
|
|
* even try to deregister from the daemon - there
|
|
* is no guarantee we won't just hang in
|
|
* the communication
|
|
*/
|
|
return ORTE_SUCCESS;
|
|
}
|
|
|
|
OPAL_OUTPUT_VERBOSE((5, orte_routed_base_framework.framework_output,
|
|
"%s %s with daemon %s",
|
|
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
|
|
ORTE_NAME_PRINT(ORTE_PROC_MY_DAEMON),
|
|
setup ? "registering" : "deregistering"));
|
|
|
|
/* we need to get the oob to establish
|
|
* the connection - the oob will leave the connection "alive"
|
|
* thereafter so we can communicate readily
|
|
*/
|
|
|
|
buffer = OBJ_NEW(opal_buffer_t);
|
|
|
|
if (setup) {
|
|
/* tell the daemon to send back a nidmap */
|
|
command = ORTE_DAEMON_SYNC_WANT_NIDMAP;
|
|
if (ORTE_SUCCESS != (rc = opal_dss.pack(buffer, &command, 1, ORTE_DAEMON_CMD))) {
|
|
ORTE_ERROR_LOG(rc);
|
|
OBJ_RELEASE(buffer);
|
|
return rc;
|
|
}
|
|
|
|
/* add our contact info to the buffer so the daemon can explicitly
|
|
* store it
|
|
*/
|
|
rml_uri = orte_rml.get_contact_info();
|
|
if (ORTE_SUCCESS != (rc = opal_dss.pack(buffer, &rml_uri, 1, OPAL_STRING))) {
|
|
ORTE_ERROR_LOG(rc);
|
|
OBJ_RELEASE(buffer);
|
|
free(rml_uri);
|
|
return rc;
|
|
}
|
|
if (NULL != rml_uri) free(rml_uri);
|
|
|
|
/* tell the daemon if we are an MPI proc */
|
|
if (ORTE_PROC_IS_MPI) {
|
|
flag = 1;
|
|
} else {
|
|
flag = 0;
|
|
}
|
|
if (ORTE_SUCCESS != (rc = opal_dss.pack(buffer, &flag, 1, OPAL_UINT8))) {
|
|
ORTE_ERROR_LOG(rc);
|
|
OBJ_RELEASE(buffer);
|
|
return rc;
|
|
}
|
|
} else {
|
|
/* deregister with the daemon */
|
|
command = ORTE_DAEMON_SYNC_BY_PROC;
|
|
if (ORTE_SUCCESS != (rc = opal_dss.pack(buffer, &command, 1, ORTE_DAEMON_CMD))) {
|
|
ORTE_ERROR_LOG(rc);
|
|
OBJ_RELEASE(buffer);
|
|
return rc;
|
|
}
|
|
}
|
|
|
|
/* setup to receive the response */
|
|
sync_waiting = true;
|
|
orte_rml.recv_buffer_nb(ORTE_NAME_WILDCARD, ORTE_RML_TAG_SYNC,
|
|
ORTE_RML_NON_PERSISTENT, report_sync, &sync_waiting);
|
|
|
|
/* send the sync command to our daemon */
|
|
if (0 > (rc = orte_rml.send_buffer_nb(ORTE_PROC_MY_DAEMON, buffer,
|
|
ORTE_RML_TAG_DAEMON,
|
|
orte_rml_send_callback, NULL))) {
|
|
ORTE_ERROR_LOG(rc);
|
|
return rc;
|
|
}
|
|
|
|
OPAL_OUTPUT_VERBOSE((5, orte_routed_base_framework.framework_output,
|
|
"%s registering sync waiting for ack",
|
|
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
|
|
|
|
|
|
/* get the ack - includes the nidmap */
|
|
ORTE_WAIT_FOR_COMPLETION(sync_waiting);
|
|
OPAL_OUTPUT_VERBOSE((5, orte_routed_base_framework.framework_output,
|
|
"%s registering sync ack recvd",
|
|
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
|
|
|
|
return ORTE_SUCCESS;
|
|
}
|
|
|
|
int orte_routed_base_process_callback(orte_jobid_t job, opal_buffer_t *buffer)
|
|
{
|
|
orte_proc_t *proc;
|
|
orte_job_t *jdata;
|
|
orte_std_cntr_t cnt;
|
|
char *rml_uri;
|
|
orte_vpid_t vpid;
|
|
int rc;
|
|
|
|
/* lookup the job object for this process */
|
|
if (NULL == (jdata = orte_get_job_data_object(job))) {
|
|
/* came from a different job family - this is an error */
|
|
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
|
|
return ORTE_ERR_NOT_FOUND;
|
|
}
|
|
|
|
/* unpack the data for each entry */
|
|
cnt = 1;
|
|
while (ORTE_SUCCESS == (rc = opal_dss.unpack(buffer, &vpid, &cnt, ORTE_VPID))) {
|
|
|
|
if (ORTE_SUCCESS != (rc = opal_dss.unpack(buffer, &rml_uri, &cnt, OPAL_STRING))) {
|
|
ORTE_ERROR_LOG(rc);
|
|
continue;
|
|
}
|
|
|
|
OPAL_OUTPUT_VERBOSE((2, orte_routed_base_framework.framework_output,
|
|
"%s routed_binomial:callback got uri %s for job %s rank %s",
|
|
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
|
|
(NULL == rml_uri) ? "NULL" : rml_uri,
|
|
ORTE_JOBID_PRINT(job), ORTE_VPID_PRINT(vpid)));
|
|
|
|
if (NULL == rml_uri) {
|
|
/* should not happen */
|
|
ORTE_ERROR_LOG(ORTE_ERR_FATAL);
|
|
return ORTE_ERR_FATAL;
|
|
}
|
|
|
|
if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(jdata->procs, vpid))) {
|
|
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
|
|
continue;
|
|
}
|
|
|
|
/* update the record */
|
|
proc->rml_uri = strdup(rml_uri);
|
|
free(rml_uri);
|
|
|
|
cnt = 1;
|
|
}
|
|
if (ORTE_ERR_UNPACK_READ_PAST_END_OF_BUFFER != rc) {
|
|
ORTE_ERROR_LOG(rc);
|
|
return rc;
|
|
}
|
|
|
|
return ORTE_SUCCESS;
|
|
}
|