Skip to content
This repository has been archived by the owner on Dec 2, 2021. It is now read-only.

enable psm2 nameserver #307

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

enable psm2 nameserver #307

wants to merge 1 commit into from

Conversation

yulujia
Copy link
Contributor

@yulujia yulujia commented Dec 12, 2019

let psm2 use ip:port addr format and use nameserver to resolve it to the
psm2 native addr

Copy link
Collaborator

@daosbuild1 daosbuild1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@@ -85,6 +85,7 @@ struct host {
};

static int my_rank;
volatile static int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise statics to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt
(style) storage class should be at the beginning of the declaration

@@ -371,6 +371,9 @@ tc_cli_start_basic(char *local_group_name, char *srv_group_name,
uint32_t grp_size;
int attach_retries = opts.num_attach_retries;
int rc = 0;
int myflag = 0;
while (myflag)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) Missing a blank line after declarations


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
// crt_na_ofi_conf.noc_ip_str, port);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@daosbuild1
Copy link
Collaborator

@daosbuild1
Copy link
Collaborator

Test stage Build on CentOS 7 with Clang completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/cart/view/change-requests/job/PR-307/1/execution/node/258/log

@yulujia yulujia force-pushed the enable-psm2-nameserver-wip branch from 2509889 to 38749b1 Compare December 12, 2019 20:27
@daosbuild1 daosbuild1 dismissed their stale review December 12, 2019 20:27

Updated patch

Copy link
Collaborator

@daosbuild1 daosbuild1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@@ -85,6 +85,7 @@ struct host {
};

static int my_rank;
volatile static int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise statics to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt
(style) storage class should be at the beginning of the declaration

@@ -371,6 +371,9 @@ tc_cli_start_basic(char *local_group_name, char *srv_group_name,
uint32_t grp_size;
int attach_retries = opts.num_attach_retries;
int rc = 0;
int myflag = 0;
while (myflag)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) Missing a blank line after declarations


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
// crt_na_ofi_conf.noc_ip_str, port);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@daosbuild1
Copy link
Collaborator

@daosbuild1
Copy link
Collaborator

Test stage Build on CentOS 7 with Clang completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/cart/view/change-requests/job/PR-307/2/execution/node/262/log

@yulujia yulujia force-pushed the enable-psm2-nameserver-wip branch from 38749b1 to 48f7d46 Compare December 12, 2019 21:43
@daosbuild1 daosbuild1 dismissed their stale review December 12, 2019 21:43

Updated patch

Copy link
Collaborator

@daosbuild1 daosbuild1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@@ -85,6 +85,7 @@ struct host {
};

static int my_rank;
volatile static int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise statics to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt
(style) storage class should be at the beginning of the declaration

@@ -371,6 +371,9 @@ tc_cli_start_basic(char *local_group_name, char *srv_group_name,
uint32_t grp_size;
int attach_retries = opts.num_attach_retries;
int rc = 0;
int myflag = 0;
while (myflag)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) Missing a blank line after declarations


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
// crt_na_ofi_conf.noc_ip_str, port);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@daosbuild1
Copy link
Collaborator

@daosbuild1
Copy link
Collaborator

Test stage Build on CentOS 7 with Clang completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/cart/view/change-requests/job/PR-307/3/execution/node/254/log

@yulujia yulujia force-pushed the enable-psm2-nameserver-wip branch from 48f7d46 to 10f0146 Compare December 12, 2019 22:16
@daosbuild1 daosbuild1 dismissed their stale review December 12, 2019 22:17

Updated patch

Copy link
Collaborator

@daosbuild1 daosbuild1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@@ -85,6 +85,7 @@ struct host {
};

static int my_rank;
volatile static int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise statics to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt
(style) storage class should be at the beginning of the declaration

@@ -371,6 +371,9 @@ tc_cli_start_basic(char *local_group_name, char *srv_group_name,
uint32_t grp_size;
int attach_retries = opts.num_attach_retries;
int rc = 0;
int myflag = 0;
while (myflag)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) Missing a blank line after declarations


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
// crt_na_ofi_conf.noc_ip_str, port);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@daosbuild1
Copy link
Collaborator

@daosbuild1
Copy link
Collaborator

Test stage Build on CentOS 7 with Clang completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/cart/view/change-requests/job/PR-307/4/execution/node/251/log

@yulujia yulujia force-pushed the enable-psm2-nameserver-wip branch from 10f0146 to aa6d98c Compare December 13, 2019 00:48
@daosbuild1 daosbuild1 dismissed their stale review December 13, 2019 00:49

Updated patch

Copy link
Collaborator

@daosbuild1 daosbuild1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@@ -85,6 +85,7 @@ struct host {
};

static int my_rank;
volatile static int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise statics to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt
(style) storage class should be at the beginning of the declaration

@@ -44,6 +44,7 @@
#include <cart/api.h>

#include "crt_internal.h"
volatile int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise globals to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
// crt_na_ofi_conf.noc_ip_str, port);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@daosbuild1
Copy link
Collaborator

@daosbuild1
Copy link
Collaborator

Test stage Build on CentOS 7 with Clang completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/cart/view/change-requests/job/PR-307/5/execution/node/266/log

@yulujia yulujia force-pushed the enable-psm2-nameserver-wip branch from aa6d98c to 3d96f55 Compare December 16, 2019 18:34
@daosbuild1 daosbuild1 dismissed their stale review December 16, 2019 18:41

Updated patch

Copy link
Collaborator

@daosbuild1 daosbuild1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@@ -85,6 +85,7 @@ struct host {
};

static int my_rank;
volatile static int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise statics to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt
(style) storage class should be at the beginning of the declaration

@@ -545,8 +545,8 @@ crt_finalize(void)
crt_gdata.cg_inited = 0;
gdata_init_flag = 0;

if (crt_gdata.cg_na_plugin == CRT_NA_OFI_SOCKETS)
crt_na_ofi_config_fini();
// if (crt_gdata.cg_na_plugin == CRT_NA_OFI_SOCKETS)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@@ -44,6 +44,7 @@
#include <cart/api.h>

#include "crt_internal.h"
volatile int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise globals to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
// crt_na_ofi_conf.noc_ip_str, port);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@daosbuild1
Copy link
Collaborator

@daosbuild1
Copy link
Collaborator

Test stage Build on CentOS 7 with Clang completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/cart/view/change-requests/job/PR-307/6/execution/node/275/log

let psm2 use ip:port addr format and use nameserver to resolve it to the
psm2 native addr. This patch is based off the cart commit used by daos
master as of 2019-12-11.

note:
1) orterun needs to use this flag:
	--mca mtl ^psm2,ofi

2) server side needs to pass this shell variable under orterun:
	-x FI_PSM2_NAME_SERVER=1
   this variable tells psm2 to start the name server

3) optionally, a OFI_PORT variable can be specified on either the server
side or client side. If OFI_PORT is not specified, cart will pick a
unused port on its own.

4) to make a server use the same address across restarts, supply this
variable when launching the server:
	-x OFI_PORT=xxx

Signed-off-by: Yulu Jia <yulu.jia@intel.com>
@yulujia yulujia force-pushed the enable-psm2-nameserver-wip branch from 3d96f55 to a6ea906 Compare December 16, 2019 23:09
@daosbuild1 daosbuild1 dismissed their stale review December 16, 2019 23:10

Updated patch

Copy link
Collaborator

@daosbuild1 daosbuild1 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@@ -85,6 +85,7 @@ struct host {
};

static int my_rank;
volatile static int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise statics to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt
(style) storage class should be at the beginning of the declaration

@@ -545,8 +545,8 @@ crt_finalize(void)
crt_gdata.cg_inited = 0;
gdata_init_flag = 0;

if (crt_gdata.cg_na_plugin == CRT_NA_OFI_SOCKETS)
crt_na_ofi_config_fini();
// if (crt_gdata.cg_na_plugin == CRT_NA_OFI_SOCKETS)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@@ -44,6 +44,7 @@
#include <cart/api.h>

#include "crt_internal.h"
volatile int myflag = 0;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not initialise globals to 0
(style) Use of volatile is usually wrong: see Documentation/volatile-considered-harmful.txt


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments


D_ASPRINTF(*string, "%s://%s/%s:%d", plugin_str,
crt_na_ofi_conf.noc_domain,
crt_na_ofi_conf.noc_ip_str, port);
// D_ASPRINTF(*string, "%s://%s:%d", plugin_str,
// crt_na_ofi_conf.noc_ip_str, port);
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(style) do not use C99 // comments

@daosbuild1
Copy link
Collaborator

@daosbuild1
Copy link
Collaborator

Test stage Build on CentOS 7 with Clang completed with status FAILURE. https://build.hpdd.intel.com//job/daos-stack/job/cart/view/change-requests/job/PR-307/7/execution/node/239/log

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants