wassname/ray
1
0
Fork 0
mirror of https://github.com/wassname/ray.git synced 2026-07-19 11:27:32 +08:00
Code Issues Packages Projects Releases Wiki Activity

Shard Redis. (#539)

Browse Source 
* Implement sharding in the Ray core

* Single node Python modifications to do sharding

* Do the sharding in redis.cc

* Pipe num_redis_shards through start_ray.py and worker.py.

* Use multiple redis shards in multinode tests.

* first steps for sharding ray.global_state

* Fix problem in multinode docker test.

* fix runtest.py

* fix some tests

* fix redis shard startup

* fix redis sharding

* fix

* fix bug introduced by the map-iterator being consumed

* fix sharding bug

* shard event table

* update number of Redis clients to be 64K

* Fix object table tests by flushing shards in between unit tests

* Fix local scheduler tests

* Documentation

* Register shard locations in the primary shard

* Add plasma unit tests back to build

* lint

* lint and fix build

* Fix

* Address Robert's comments

* Refactor start_ray_processes to start Redis shard

* lint

* Fix global scheduler python tests

* Fix redis module test

* Fix plasma test

* Fix component failure test

* Fix local scheduler test

* Fix runtest.py

* Fix global scheduler test for python3

* Fix task_table_test_and_update bug, from actor task table submission race

* Fix jenkins tests.

* Retry Redis shard connections

* Fix test cases

* Convert database clients to DBClient struct

* Fix race condition when subscribing to db client table

* Remove unused lines, add APITest for sharded Ray

* Fix

* Fix memory leak

* Suppress ReconstructionTests output

* Suppress output for APITestSharded

* Reissue task table add/update commands if initial command does not publish to any subscribers.

* fix

* Fix linting.

* fix tests

* fix linting

* fix python test

* fix linting
This commit is contained in:
Stephanie Wang
2017-05-18 17:40:41 -07:00
committed by Philipp Moritz
parent 0a4304725f
commit ee08c8274b
39 changed files with 1336 additions and 651 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/common/test/task_table_tests.cc
+47 -22
View File
@@ -40,8 +40,8 @@ void lookup_nil_success_callback(Task *task, void *context) {
TEST lookup_nil_test(void) {
lookup_nil_id = globally_unique_id();
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
RetryInfo retry = {
.num_retries = 5,
@@ -96,14 +96,16 @@ void add_success_callback(TaskID task_id, void *context) {
TEST add_lookup_test(void) {
add_lookup_task = example_task(1, 1, TASK_STATUS_WAITING);
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
RetryInfo retry = {
.num_retries = 5,
.timeout = 1000,
.fail_callback = add_lookup_fail_callback,
};
task_table_subscribe(db, NIL_ID, TASK_STATUS_WAITING, NULL, NULL, &retry,
NULL, NULL);
task_table_add_task(db, Task_copy(add_lookup_task), &retry,
add_success_callback, (void *) db);
/* Disconnect the database to see if the lookup times out. */
@@ -136,8 +138,8 @@ void subscribe_fail_callback(UniqueID id, void *user_context, void *user_data) {
TEST subscribe_timeout_test(void) {
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
RetryInfo retry = {
.num_retries = 5,
@@ -148,7 +150,10 @@ TEST subscribe_timeout_test(void) {
subscribe_done_callback,
(void *) subscribe_timeout_context);
/* Disconnect the database to see if the subscribe times out. */
close(db->sub_context->c.fd);
close(db->subscribe_context->c.fd);
for (int i = 0; i < db->subscribe_contexts.size(); ++i) {
close(db->subscribe_contexts[i]->c.fd);
}
aeProcessEvents(g_loop, AE_TIME_EVENTS);
event_loop_run(g_loop);
db_disconnect(db);
@@ -177,17 +182,22 @@ void publish_fail_callback(UniqueID id, void *user_context, void *user_data) {
TEST publish_timeout_test(void) {
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
Task *task = example_task(1, 1, TASK_STATUS_WAITING);
RetryInfo retry = {
.num_retries = 5, .timeout = 100, .fail_callback = publish_fail_callback,
};
task_table_subscribe(db, NIL_ID, TASK_STATUS_WAITING, NULL, NULL, &retry,
NULL, NULL);
task_table_add_task(db, task, &retry, publish_done_callback,
(void *) publish_timeout_context);
/* Disconnect the database to see if the publish times out. */
close(db->context->c.fd);
for (int i = 0; i < db->contexts.size(); ++i) {
close(db->contexts[i]->c.fd);
}
aeProcessEvents(g_loop, AE_TIME_EVENTS);
event_loop_run(g_loop);
db_disconnect(db);
@@ -204,9 +214,14 @@ int64_t reconnect_db_callback(event_loop *loop,
void *context) {
DBHandle *db = (DBHandle *) context;
/* Reconnect to redis. */
redisAsyncFree(db->sub_context);
db->sub_context = redisAsyncConnect("127.0.0.1", 6379);
db->sub_context->data = (void *) db;
redisAsyncFree(db->subscribe_context);
db->subscribe_context = redisAsyncConnect("127.0.0.1", 6379);
db->subscribe_context->data = (void *) db;
for (int i = 0; i < db->subscribe_contexts.size(); ++i) {
redisAsyncFree(db->subscribe_contexts[i]);
db->subscribe_contexts[i] = redisAsyncConnect("127.0.0.1", 6380 + i);
db->subscribe_contexts[i]->data = (void *) db;
}
/* Re-attach the database to the event loop (the file descriptor changed). */
db_attach(db, loop, true);
return EVENT_LOOP_TIMER_DONE;
@@ -239,8 +254,8 @@ void subscribe_retry_fail_callback(UniqueID id,
TEST subscribe_retry_test(void) {
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
RetryInfo retry = {
.num_retries = 5,
@@ -251,7 +266,10 @@ TEST subscribe_retry_test(void) {
subscribe_retry_done_callback,
(void *) subscribe_retry_context);
/* Disconnect the database to see if the subscribe times out. */
close(db->sub_context->c.fd);
close(db->subscribe_context->c.fd);
for (int i = 0; i < db->subscribe_contexts.size(); ++i) {
close(db->subscribe_contexts[i]->c.fd);
}
/* Install handler for reconnecting the database. */
event_loop_add_timer(g_loop, 150,
(event_loop_timer_handler) reconnect_db_callback, db);
@@ -286,8 +304,8 @@ void publish_retry_fail_callback(UniqueID id,
TEST publish_retry_test(void) {
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
Task *task = example_task(1, 1, TASK_STATUS_WAITING);
RetryInfo retry = {
@@ -295,10 +313,15 @@ TEST publish_retry_test(void) {
.timeout = 100,
.fail_callback = publish_retry_fail_callback,
};
task_table_subscribe(db, NIL_ID, TASK_STATUS_WAITING, NULL, NULL, &retry,
NULL, NULL);
task_table_add_task(db, task, &retry, publish_retry_done_callback,
(void *) publish_retry_context);
/* Disconnect the database to see if the publish times out. */
close(db->sub_context->c.fd);
close(db->subscribe_context->c.fd);
for (int i = 0; i < db->subscribe_contexts.size(); ++i) {
close(db->subscribe_contexts[i]->c.fd);
}
/* Install handler for reconnecting the database. */
event_loop_add_timer(g_loop, 150,
(event_loop_timer_handler) reconnect_db_callback, db);
@@ -335,8 +358,8 @@ void subscribe_late_done_callback(TaskID task_id, void *user_context) {
TEST subscribe_late_test(void) {
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
RetryInfo retry = {
.num_retries = 0,
@@ -380,8 +403,8 @@ void publish_late_done_callback(TaskID task_id, void *user_context) {
TEST publish_late_test(void) {
g_loop = event_loop_create();
DBHandle *db =
db_connect("127.0.0.1", 6379, "plasma_manager", "127.0.0.1", 0, NULL);
DBHandle *db = db_connect(std::string("127.0.0.1"), 6379, "plasma_manager",
"127.0.0.1", 0, NULL);
db_attach(db, g_loop, false);
Task *task = example_task(1, 1, TASK_STATUS_WAITING);
RetryInfo retry = {
@@ -389,6 +412,8 @@ TEST publish_late_test(void) {
.timeout = 0,
.fail_callback = publish_late_fail_callback,
};
task_table_subscribe(db, NIL_ID, TASK_STATUS_WAITING, NULL, NULL, NULL, NULL,
NULL);
task_table_add_task(db, task, &retry, publish_late_done_callback,
(void *) publish_late_context);
/* Install handler for terminating the event loop. */