54b2cf747e
The commit has been tested for C/R and Cray operations, and on Odin (SLURM, rsh) and RoadRunner (TM). I tried to update all environments, but obviously could not test them. I know that Windows needs some work, and have highlighted what is know to be needed in the odls process component. This represents a lot of work by Brian, Tim P, Josh, and myself, with much advice from Jeff and others. For posterity, I have appended a copy of the email describing the work that was done: As we have repeatedly noted, the modex operation in MPI_Init is the single greatest consumer of time during startup. To-date, we have executed that operation as an ORTE stage gate that held the process until a startup message containing all required modex (and OOB contact info - see #3 below) info could be sent to it. Each process would send its data to the HNP's registry, which assembled and sent the message when all processes had reported in. In addition, ORTE had taken responsibility for monitoring process status as it progressed through a series of "stage gates". The process reported its status at each gate, and ORTE would then send a "release" message once all procs had reported in. The incoming changes revamp these procedures in three ways: 1. eliminating the ORTE stage gate system and cleanly delineating responsibility between the OMPI and ORTE layers for MPI init/finalize. The modex stage gate (STG1) has been replaced by a collective operation in the modex itself that performs an allgather on the required modex info. The allgather is implemented using the orte_grpcomm framework since the BTL's are not active at that point. At the moment, the grpcomm framework only has a "basic" component analogous to OMPI's "basic" coll framework - I would recommend that the MPI team create additional, more advanced components to improve performance of this step. The other stage gates have been replaced by orte_grpcomm barrier functions. We tried to use MPI barriers instead (since the BTL's are active at that point), but - as we discussed on the telecon - these are not currently true barriers so the job would hang when we fell through while messages were still in process. Note that the grpcomm barrier doesn't actually resolve that problem, but Brian has pointed out that we are unlikely to ever see it violated. Again, you might want to spend a little time on an advanced barrier algorithm as the one in "basic" is very simplistic. Summarizing this change: ORTE no longer tracks process state nor has direct responsibility for synchronizing jobs. This is now done via collective operations within the MPI layer, albeit using ORTE collective communication services. I -strongly- urge the MPI team to implement advanced collective algorithms to improve the performance of this critical procedure. 2. reducing the volume of data exchanged during modex. Data in the modex consisted of the process name, the name of the node where that process is located (expressed as a string), plus a string representation of all contact info. The nodename was required in order for the modex to determine if the process was local or not - in addition, some people like to have it to print pretty error messages when a connection failed. The size of this data has been reduced in three ways: (a) reducing the size of the process name itself. The process name consisted of two 32-bit fields for the jobid and vpid. This is far larger than any current system, or system likely to exist in the near future, can support. Accordingly, the default size of these fields has been reduced to 16-bits, which means you can have 32k procs in each of 32k jobs. Since the daemons must have a vpid, and we require one daemon/node, this also restricts the default configuration to 32k nodes. To support any future "mega-clusters", a configuration option --enable-jumbo-apps has been added. This option increases the jobid and vpid field sizes to 32-bits. Someday, if necessary, someone can add yet another option to increase them to 64-bits, I suppose. (b) replacing the string nodename with an integer nodeid. Since we have one daemon/node, the nodeid corresponds to the local daemon's vpid. This replaces an often lengthy string with only 2 (or at most 4) bytes, a substantial reduction. (c) when the mca param requesting that nodenames be sent to support pretty error messages, a second mca param is now used to request FQDN - otherwise, the domain name is stripped (by default) from the message to save space. If someone wants to combine those into a single param somehow (perhaps with an argument?), they are welcome to do so - I didn't want to alter what people are already using. While these may seem like small savings, they actually amount to a significant impact when aggregated across the entire modex operation. Since every proc must receive the modex data regardless of the collective used to send it, just reducing the size of the process name removes nearly 400MBytes of communication from a 32k proc job (admittedly, much of this comm may occur in parallel). So it does add up pretty quickly. 3. routing RML messages to reduce connections. The default messaging system remains point-to-point - i.e., each proc opens a socket to every proc it communicates with and sends its messages directly. A new option uses the orteds as routers - i.e., each proc only opens a single socket to its local orted. All messages are sent from the proc to the orted, which forwards the message to the orted on the node where the intended recipient proc is located - that orted then forwards the message to its local proc (the recipient). This greatly reduces the connection storm we have encountered during startup. It also has the benefit of removing the sharing of every proc's OOB contact with every other proc. The orted routing tables are populated during launch since every orted gets a map of where every proc is being placed. Each proc, therefore, only needs to know the contact info for its local daemon, which is passed in via the environment when the proc is fork/exec'd by the daemon. This alone removes ~50 bytes/process of communication that was in the current STG1 startup message - so for our 32k proc job, this saves us roughly 32k*50 = 1.6MBytes sent to 32k procs = 51GBytes of messaging. Note that you can use the new routing method by specifying -mca routed tree - if you so desire. This mode will become the default at some point in the future. There are a few minor additional changes in the commit that I'll just note in passing: * propagation of command line mca params to the orteds - fixes ticket #1073. See note there for details. * requiring of "finalize" prior to "exit" for MPI procs - fixes ticket #1144. See note there for details. * cleanup of some stale header files This commit was SVN r16364.
235 строки
8.1 KiB
C
235 строки
8.1 KiB
C
/*
|
|
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
|
|
* University Research and Technology
|
|
* Corporation. All rights reserved.
|
|
* Copyright (c) 2004-2006 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.
|
|
* $COPYRIGHT$
|
|
*
|
|
* Additional copyrights may follow
|
|
*
|
|
* $HEADER$
|
|
*/
|
|
/** @file:
|
|
*
|
|
* The Open MPI State-of-Health Monitoring Subsystem
|
|
*
|
|
*/
|
|
|
|
#ifndef ORTE_SMR_H
|
|
#define ORTE_SMR_H
|
|
|
|
/*
|
|
* includes
|
|
*/
|
|
|
|
#include "orte_config.h"
|
|
#include "orte/orte_constants.h"
|
|
#include "orte/orte_types.h"
|
|
|
|
#include "opal/mca/mca.h"
|
|
|
|
#include "orte/mca/gpr/gpr_types.h"
|
|
#include "orte/mca/ns/ns_types.h"
|
|
#include "orte/mca/smr/smr_types.h"
|
|
#include "orte/mca/rmaps/rmaps_types.h"
|
|
|
|
#if defined(c_plusplus) || defined(__cplusplus)
|
|
extern "C" {
|
|
#endif
|
|
|
|
/*
|
|
* Component functions - all MUST be provided!
|
|
*/
|
|
|
|
|
|
/*
|
|
* Query a process state
|
|
*/
|
|
typedef int (*orte_smr_base_module_get_proc_state_fn_t)(orte_proc_state_t *state,
|
|
int *status,
|
|
orte_process_name_t *proc);
|
|
|
|
/*
|
|
* Set a process state
|
|
*/
|
|
typedef int (*orte_smr_base_module_set_proc_state_fn_t)(orte_process_name_t *proc,
|
|
orte_proc_state_t state, int status);
|
|
|
|
/*
|
|
* Query a node state
|
|
*/
|
|
typedef int (*orte_smr_base_module_get_node_state_fn_t)(orte_node_state_t *state,
|
|
char *nodename);
|
|
/*
|
|
* Set a node state
|
|
*/
|
|
typedef int (*orte_smr_base_module_set_node_state_fn_t)(char *nodename,
|
|
orte_node_state_t state);
|
|
|
|
/*
|
|
* Query a job state
|
|
*/
|
|
typedef int (*orte_smr_base_module_get_job_state_fn_t)(orte_job_state_t *state,
|
|
orte_jobid_t jobid);
|
|
|
|
/*
|
|
* Set a job state
|
|
*/
|
|
typedef int (*orte_smr_base_module_set_job_state_fn_t)(orte_jobid_t jobid,
|
|
orte_job_state_t state);
|
|
|
|
/*
|
|
* Register a sync request
|
|
*/
|
|
typedef int (*orte_smr_base_module_register_sync_fn_t)(void);
|
|
|
|
/*
|
|
* Define the job-specific standard stage gates
|
|
* This function creates all of the ORTE-standard stage gates.
|
|
*/
|
|
typedef int (*orte_smr_base_module_job_stage_gate_init_fn_t)(orte_jobid_t job,
|
|
orte_gpr_trigger_cb_fn_t cbfunc,
|
|
void *user_tag);
|
|
|
|
/*
|
|
* Define an "alert" monitor
|
|
* This function will establish an appropriate trigger to notify the specified
|
|
* callback function when an event takes place. In this case, event is defined
|
|
* by the specified memory location achieving the specified value - e.g., a
|
|
* location could be monitored for a value being set to 1, indicating that a
|
|
* process has aborted.
|
|
*
|
|
* @param job The job that is to be monitored.
|
|
*
|
|
* @param *trigger_name The name of the trigger to be defined.
|
|
*
|
|
* @param *counter_key A string defining the key name of the counter on the registry.
|
|
*
|
|
* @param *counter A pointer to a data_value object that contains the initial
|
|
* value to which the counter should be set.
|
|
*
|
|
* @param *alert_value A pointer to a data_value object that contains the value of
|
|
* the counter that should cause the alert to be sent.
|
|
*
|
|
* @param one_shot Whether or not the trigger should be a one-shot
|
|
*
|
|
* @param cbfunc A registry callback function to be called when the alert fires.
|
|
*
|
|
* @param *user_tag Whatever data the user would like to have passed back to them
|
|
* when the alert is received
|
|
*
|
|
* NOTE: alerts are intended solely for purposes of alerting the caller when
|
|
* an event happens. Thus, they do not convey any information beyond the fact that
|
|
* they fired.
|
|
*/
|
|
typedef int (*orte_smr_base_module_define_alert_monitor_fn_t)(orte_jobid_t job,
|
|
char *trigger_name,
|
|
char *counter_key,
|
|
orte_std_cntr_t counter,
|
|
orte_std_cntr_t alert_value,
|
|
bool one_shot,
|
|
orte_gpr_trigger_cb_fn_t cbfunc,
|
|
void *user_tag);
|
|
|
|
/*
|
|
* Initiate monitoring of a job
|
|
* This function notifies the smr that it should initiate monitoring of the specified
|
|
* jobid. It is called by a PLS component at an appropriate point in the launch procedure. Calling
|
|
* the function allows smr components (e.g., the BProc component that monitors daemons
|
|
* via the BProc-provided centralized alerting system) to make the necessary connections
|
|
* for monitoring the job.
|
|
*/
|
|
typedef int (*orte_smr_base_module_begin_monitoring_fn_t)(orte_job_map_t *map,
|
|
orte_gpr_trigger_cb_fn_t cbfunc,
|
|
void *user_tag);
|
|
|
|
/*
|
|
* Subscribe to a job stage gate
|
|
*/
|
|
typedef int (*orte_smr_base_module_job_stage_gate_subscribe_fn_t)(orte_jobid_t job,
|
|
orte_gpr_notify_cb_fn_t cbfunc, void* cbdata,
|
|
orte_proc_state_t cb_conditions);
|
|
|
|
|
|
/* Shutdown the module nicely
|
|
*/
|
|
|
|
typedef int (*orte_smr_base_module_finalize_fn_t)(void);
|
|
|
|
|
|
|
|
/* below are the prototypes needed by the MCA */
|
|
|
|
/*
|
|
* Ver 1.3.0
|
|
*/
|
|
struct orte_smr_base_module_1_3_0_t {
|
|
orte_smr_base_module_get_proc_state_fn_t get_proc_state;
|
|
orte_smr_base_module_set_proc_state_fn_t set_proc_state;
|
|
orte_smr_base_module_get_node_state_fn_t get_node_state;
|
|
orte_smr_base_module_set_node_state_fn_t set_node_state;
|
|
orte_smr_base_module_get_job_state_fn_t get_job_state;
|
|
orte_smr_base_module_set_job_state_fn_t set_job_state;
|
|
orte_smr_base_module_begin_monitoring_fn_t begin_monitoring;
|
|
/* TRIGGER INIT FUNCTIONS */
|
|
orte_smr_base_module_job_stage_gate_init_fn_t init_job_stage_gates;
|
|
orte_smr_base_module_define_alert_monitor_fn_t define_alert_monitor;
|
|
orte_smr_base_module_job_stage_gate_subscribe_fn_t job_stage_gate_subscribe;
|
|
/* REGISTER FUNCTION */
|
|
orte_smr_base_module_register_sync_fn_t register_sync;
|
|
/* FINALIZE */
|
|
orte_smr_base_module_finalize_fn_t finalize;
|
|
};
|
|
|
|
typedef struct orte_smr_base_module_1_3_0_t orte_smr_base_module_1_3_0_t;
|
|
typedef orte_smr_base_module_1_3_0_t orte_smr_base_module_t;
|
|
|
|
/*
|
|
* SOH Component
|
|
*/
|
|
|
|
typedef orte_smr_base_module_t* (*orte_smr_base_component_init_fn_t)(
|
|
int *priority);
|
|
|
|
typedef int (*orte_smr_base_component_finalize_fn_t)(void);
|
|
|
|
/*
|
|
* the standard component data structure
|
|
*/
|
|
|
|
struct orte_smr_base_component_1_3_0_t {
|
|
mca_base_component_t smr_version;
|
|
mca_base_component_data_1_0_0_t smr_data;
|
|
orte_smr_base_component_init_fn_t smr_init;
|
|
orte_smr_base_component_finalize_fn_t smr_finalize;
|
|
};
|
|
|
|
typedef struct orte_smr_base_component_1_3_0_t orte_smr_base_component_1_3_0_t;
|
|
|
|
typedef orte_smr_base_component_1_3_0_t orte_smr_base_component_t;
|
|
|
|
|
|
|
|
/*
|
|
* Macro for use in components that are of type ns v1.0.0
|
|
*/
|
|
#define ORTE_SMR_BASE_VERSION_1_3_0 \
|
|
/* smr v1.3 is chained to MCA v1.0 */ \
|
|
MCA_BASE_VERSION_1_0_0, \
|
|
/* smr v1.3 */ \
|
|
"smr", 1, 3, 0
|
|
|
|
ORTE_DECLSPEC extern orte_smr_base_module_t orte_smr; /* holds selected module's function pointers */
|
|
|
|
#if defined(c_plusplus) || defined(__cplusplus)
|
|
}
|
|
#endif
|
|
|
|
#endif /* ORTE_SMR_H */
|