Skip to content
Merged
Changes from 1 commit
Commits
Show all changes
76 commits
Select commit Hold shift + click to select a range
711356b
[SPARK-3086] [SPARK-3043] [SPARK-3156] [mllib] DecisionTree aggregat…
jkbradley Sep 8, 2014
e16a8e7
SPARK-3337 Paranoid quoting in shell to allow install dirs with space…
ScrapCodes Sep 8, 2014
16a73c2
SPARK-2978. Transformation with MR shuffle semantics
sryza Sep 8, 2014
386bc24
Provide a default PYSPARK_PYTHON for python/run_tests
Sep 8, 2014
26bc765
[SQL] Minor edits to sql programming guide.
hcook Sep 8, 2014
939a322
[SPARK-3417] Use new-style classes in PySpark
mrocklin Sep 8, 2014
08ce188
[SPARK-3019] Pluggable block transfer interface (BlockTransferService)
rxin Sep 8, 2014
7db5339
[SPARK-3349][SQL] Output partitioning of limit should not be inherite…
Sep 8, 2014
50a4fa7
[SPARK-3443][MLLIB] update default values of tree:
mengxr Sep 9, 2014
ca0348e
SPARK-3423: [SQL] Implement BETWEEN for SQLParser
willb Sep 9, 2014
dc1dbf2
[SPARK-3414][SQL] Stores analyzed logical plan when registering a tem…
liancheng Sep 9, 2014
2b7ab81
[SPARK-3329][SQL] Don't depend on Hive SET pair ordering in tests.
willb Sep 9, 2014
092e2f1
SPARK-2425 Don't kill a still-running Application because of some mis…
markhamstra Sep 9, 2014
ce5cb32
[Build] Removed -Phive-thriftserver since this profile has been removed
liancheng Sep 9, 2014
c419e4f
[Docs] actorStream storageLevel default is MEMORY_AND_DISK_SER_2
melrief Sep 9, 2014
1e03cf7
[SPARK-3455] [SQL] **HOT FIX** Fix the unit test failure
chenghao-intel Sep 9, 2014
88547a0
SPARK-3422. JavaAPISuite.getHadoopInputSplits isn't used anywhere.
sryza Sep 9, 2014
f0f1ba0
SPARK-3404 [BUILD] SparkSubmitSuite fails with "spark-submit exits wi…
srowen Sep 9, 2014
2686233
[SPARK-3193]output errer info when Process exit code is not zero in t…
scwf Sep 9, 2014
02b5ac7
Minor - Fix trivial compilation warnings.
ScrapCodes Sep 9, 2014
07ee4a2
[SPARK-3176] Implement 'ABS and 'LAST' for sql
Sep 9, 2014
c110614
[SPARK-3448][SQL] Check for null in SpecificMutableRow.update
liancheng Sep 10, 2014
25b5b86
[SPARK-3458] enable python "with" statements for SparkContext
Sep 10, 2014
b734ed0
[SPARK-3395] [SQL] DSL sometimes incorrectly reuses attribute ids, br…
Sep 10, 2014
6f7a768
[SPARK-3286] - Cannot view ApplicationMaster UI when Yarn’s url schem…
Sep 10, 2014
a028330
[SPARK-3362][SQL] Fix resolution for casewhen with nulls.
adrian-wang Sep 10, 2014
f0c87dc
[SPARK-3363][SQL] Type Coercion should promote null to all other types.
adrian-wang Sep 10, 2014
26503fd
[HOTFIX] Fix scala style issue introduced by #2276.
JoshRosen Sep 10, 2014
1f4a648
SPARK-1713. Use a thread pool for launching executors.
sryza Sep 10, 2014
e4f4886
[SPARK-2096][SQL] Correctly parse dot notations
cloud-fan Sep 10, 2014
558962a
[SPARK-3411] Improve load-balancing of concurrently-submitted drivers…
WangTaoTheTonic Sep 10, 2014
79cdb9b
[SPARK-2207][SPARK-3272][MLLib]Add minimum information gain and minim…
Sep 10, 2014
84e2c8b
[SQL] Add test case with workaround for reading partitioned Avro files
marmbrus Sep 11, 2014
f92cde2
[SPARK-3447][SQL] Remove explicit conversion with JListWrapper to avo…
marmbrus Sep 11, 2014
c27718f
[SPARK-2781][SQL] Check resolution of LogicalPlans in Analyzer.
staple Sep 11, 2014
ed1980f
[SPARK-2140] Updating heap memory calculation for YARN stable and alpha.
Sep 11, 2014
1ef656e
[SPARK-3047] [PySpark] add an option to use str in textFileRDD
davies Sep 11, 2014
ca83f1e
[SPARK-2917] [SQL] Avoid table creation in logical plan analyzing for…
chenghao-intel Sep 11, 2014
4bc9e04
[SPARK-3390][SQL] sqlContext.jsonRDD fails on a complex structure of …
yhuai Sep 11, 2014
6324eb7
[Spark-3490] Disable SparkUI for tests
andrewor14 Sep 12, 2014
ce59725
[SPARK-3429] Don't include the empty string "" as a defaultAclUser
ash211 Sep 12, 2014
f858f46
SPARK-3462 push down filters and projections into Unions
Sep 12, 2014
33c7a73
SPARK-2482: Resolve sbt warnings during build
witgo Sep 12, 2014
42904b8
[SPARK-3465] fix task metrics aggregation in local mode
davies Sep 12, 2014
b8634df
[SPARK-3160] [SPARK-3494] [mllib] DecisionTree: eliminate pre-alloca…
jkbradley Sep 12, 2014
f116f76
[SPARK-2558][DOCS] Add --queue example to YARN doc
kramimus Sep 12, 2014
5333776
[PySpark] Add blank line so that Python RDD.top() docstring renders c…
rnowling Sep 12, 2014
8194fc6
[SPARK-3481] [SQL] Eliminate the error log in local Hive comparison test
chenghao-intel Sep 12, 2014
eae81b0
MAINTENANCE: Automated closing of pull requests.
pwendell Sep 12, 2014
15a5645
[SPARK-3427] [GraphX] Avoid active vertex tracking in static PageRank
ankurdave Sep 12, 2014
1d76796
SPARK-3014. Log a more informative messages in a couple failure scena…
sryza Sep 12, 2014
af25838
[SPARK-3217] Add Guava to classpath when SPARK_PREPEND_CLASSES is set.
Sep 12, 2014
25311c2
[SPARK-3456] YarnAllocator on alpha can lose container requests to RM
tgravescs Sep 13, 2014
71af030
[SPARK-3094] [PySpark] compatitable with PyPy
davies Sep 13, 2014
885d162
[SPARK-3500] [SQL] use JavaSchemaRDD as SchemaRDD._jschema_rdd
davies Sep 13, 2014
6d887db
[SPARK-3515][SQL] Moves test suite setup code to beforeAll rather tha…
liancheng Sep 13, 2014
2584ea5
[SPARK-3469] Make sure all TaskCompletionListener are called even wit…
rxin Sep 13, 2014
e11eeb7
[SQL][Docs] Update SQL programming guide to show the correct default …
yhuai Sep 13, 2014
feaa370
SPARK-3470 [CORE] [STREAMING] Add Closeable / close() to Java context…
srowen Sep 13, 2014
b4dded4
Proper indent for the previous commit.
rxin Sep 13, 2014
a523cea
[SQL] [Docs] typo fixes
nchammas Sep 13, 2014
184cd51
[SPARK-3481][SQL] Removes the evil MINOR HACK
liancheng Sep 13, 2014
7404924
[SPARK-3294][SQL] Eliminates boxing costs from in-memory columnar sto…
liancheng Sep 13, 2014
0f8c4ed
[SQL] Decrease partitions when testing
marmbrus Sep 13, 2014
2aea0da
[SPARK-3030] [PySpark] Reuse Python worker
davies Sep 13, 2014
4e3fbe8
[SPARK-3463] [PySpark] aggregate and show spilled bytes in Python
davies Sep 14, 2014
c243b21
SPARK-3039: Allow spark to be built using avro-mapred for hadoop2
bbossy Sep 15, 2014
f493f79
[SPARK-3452] Maven build should skip publishing artifacts people shou…
ScrapCodes Sep 15, 2014
cc14644
[SPARK-3410] The priority of shutdownhook for ApplicationMaster shoul…
sarutak Sep 15, 2014
fe2b1d6
[SPARK-3425] do not set MaxPermSize for OpenJDK 1.8
Sep 15, 2014
e59fac1
[SPARK-3518] Remove wasted statement in JsonProtocol
sarutak Sep 15, 2014
37d9252
[SPARK-2714] DAGScheduler logs jobid when runJob finishes
YanTangZhai Sep 15, 2014
3b93128
[SPARK-3396][MLLIB] Use SquaredL2Updater in LogisticRegressionWithSGD
BigCrunsh Sep 16, 2014
983d6a9
[MLlib] Update SVD documentation in IndexedRowMatrix
rezazadeh Sep 16, 2014
fdb302f
[SPARK-3516] [mllib] DecisionTree: Add minInstancesPerNode, minInfoGa…
Sep 16, 2014
da33acb
[SPARK-2951] [PySpark] support unpickle array.array for Python 2.6
davies Sep 16, 2014
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
[SPARK-3411] Improve load-balancing of concurrently-submitted drivers…
… across workers

If the waiting driver array is too big, the drivers in it will be dispatched to the first worker we get(if it has enough resources), with or without the Randomization.

We should do randomization every time we dispatch a driver, in order to better balance drivers.

Author: WangTaoTheTonic <[email protected]>
Author: WangTao <[email protected]>

Closes apache#1106 from WangTaoTheTonic/fixBalanceDrivers and squashes the following commits:

d1a928b [WangTaoTheTonic] Minor adjustment
b6560cf [WangTaoTheTonic] solve the shuffle problem for HashSet
f674e59 [WangTaoTheTonic] add comment and minor fix
2835929 [WangTao] solve the failed test and avoid filtering
2ca3091 [WangTao] fix checkstyle
bc91bb1 [WangTao] Avoid shuffle every time we schedule the driver using round robin
bbc7087 [WangTaoTheTonic] Optimize the schedule in Master
  • Loading branch information
WangTaoTheTonic authored and andrewor14 committed Sep 10, 2014
commit 558962a83fb0758ab5c13ff4ea58cc96c29cbbcc
18 changes: 15 additions & 3 deletions core/src/main/scala/org/apache/spark/deploy/master/Master.scala
Original file line number Diff line number Diff line change
Expand Up @@ -487,13 +487,25 @@ private[spark] class Master(
if (state != RecoveryState.ALIVE) { return }

// First schedule drivers, they take strict precedence over applications
val shuffledWorkers = Random.shuffle(workers) // Randomization helps balance drivers
for (worker <- shuffledWorkers if worker.state == WorkerState.ALIVE) {
for (driver <- List(waitingDrivers: _*)) { // iterate over a copy of waitingDrivers
// Randomization helps balance drivers
val shuffledAliveWorkers = Random.shuffle(workers.toSeq.filter(_.state == WorkerState.ALIVE))
val aliveWorkerNum = shuffledAliveWorkers.size
var curPos = 0
for (driver <- waitingDrivers.toList) { // iterate over a copy of waitingDrivers
// We assign workers to each waiting driver in a round-robin fashion. For each driver, we
// start from the last worker that was assigned a driver, and continue onwards until we have
// explored all alive workers.
curPos = (curPos + 1) % aliveWorkerNum
val startPos = curPos
var launched = false
while (curPos != startPos && !launched) {
val worker = shuffledAliveWorkers(curPos)
if (worker.memoryFree >= driver.desc.mem && worker.coresFree >= driver.desc.cores) {
launchDriver(worker, driver)
waitingDrivers -= driver
launched = true
}
curPos = (curPos + 1) % aliveWorkerNum
}
}

Expand Down