Github user rxin commented on the pull request:
https://github.com/apache/spark/pull/2030#issuecomment-52675858
I tested this with a local cluster (which might not be a representative
experiment), but this commit makes dummy short tasks finish in about 1/4 of the
time.
```
rxin @ rxin-mbp : /scratch/rxin/spark-1 (torrentBroadcast)
> grep "took" new.txt
14/08/19 11:06:56 INFO SparkContext: Job finished: count at <console>:13,
took 0.169695 s
14/08/19 11:06:58 INFO SparkContext: Job finished: count at <console>:13,
took 0.010982 s
14/08/19 11:06:59 INFO SparkContext: Job finished: count at <console>:13,
took 0.010995 s
14/08/19 11:07:00 INFO SparkContext: Job finished: count at <console>:13,
took 0.009852 s
14/08/19 11:07:01 INFO SparkContext: Job finished: count at <console>:13,
took 0.007796 s
14/08/19 11:07:02 INFO SparkContext: Job finished: count at <console>:13,
took 0.009224 s
14/08/19 11:07:03 INFO SparkContext: Job finished: count at <console>:13,
took 0.007944 s
14/08/19 11:07:03 INFO SparkContext: Job finished: count at <console>:13,
took 0.011442 s
14/08/19 11:07:04 INFO SparkContext: Job finished: count at <console>:13,
took 0.007945 s
14/08/19 11:07:05 INFO SparkContext: Job finished: count at <console>:13,
took 0.008433 s
14/08/19 11:07:06 INFO SparkContext: Job finished: count at <console>:13,
took 0.008017 s
14/08/19 11:07:06 INFO SparkContext: Job finished: count at <console>:13,
took 0.0085 s
14/08/19 11:07:07 INFO SparkContext: Job finished: count at <console>:13,
took 0.008712 s
14/08/19 11:07:08 INFO SparkContext: Job finished: count at <console>:13,
took 0.008481 s
14/08/19 11:07:09 INFO SparkContext: Job finished: count at <console>:13,
took 0.007848 s
14/08/19 11:07:10 INFO SparkContext: Job finished: count at <console>:13,
took 0.008487 s
14/08/19 11:07:10 INFO SparkContext: Job finished: count at <console>:13,
took 0.00756 s
14/08/19 11:07:11 INFO SparkContext: Job finished: count at <console>:13,
took 0.008702 s
14/08/19 11:07:12 INFO SparkContext: Job finished: count at <console>:13,
took 0.007459 s
14/08/19 11:07:13 INFO SparkContext: Job finished: count at <console>:13,
took 0.007884 s
14/08/19 11:07:14 INFO SparkContext: Job finished: count at <console>:13,
took 0.007632 s
```
```
rxin @ rxin-mbp : /scratch/rxin/spark-1 (torrentBroadcast)
> grep "took" old.txt
14/08/19 11:12:12 INFO SparkContext: Job finished: count at <console>:13,
took 0.316267 s
14/08/19 11:12:14 INFO SparkContext: Job finished: count at <console>:13,
took 0.035347 s
14/08/19 11:12:15 INFO SparkContext: Job finished: count at <console>:13,
took 0.038807 s
14/08/19 11:12:16 INFO SparkContext: Job finished: count at <console>:13,
took 0.071991 s
14/08/19 11:12:16 INFO SparkContext: Job finished: count at <console>:13,
took 0.036433 s
14/08/19 11:12:17 INFO SparkContext: Job finished: count at <console>:13,
took 0.036449 s
14/08/19 11:12:29 INFO SparkContext: Job finished: count at <console>:13,
took 0.076298 s
14/08/19 11:12:29 INFO SparkContext: Job finished: count at <console>:13,
took 0.033312 s
14/08/19 11:12:30 INFO SparkContext: Job finished: count at <console>:13,
took 0.036079 s
14/08/19 11:12:31 INFO SparkContext: Job finished: count at <console>:13,
took 0.033334 s
14/08/19 11:13:15 INFO SparkContext: Job finished: count at <console>:13,
took 0.47008 s
14/08/19 11:13:17 INFO SparkContext: Job finished: count at <console>:13,
took 0.033396 s
14/08/19 11:13:18 INFO SparkContext: Job finished: count at <console>:13,
took 0.036782 s
14/08/19 11:13:19 INFO SparkContext: Job finished: count at <console>:13,
took 0.033261 s
14/08/19 11:13:19 INFO SparkContext: Job finished: count at <console>:13,
took 0.032565 s
14/08/19 11:13:20 INFO SparkContext: Job finished: count at <console>:13,
took 0.033861 s
14/08/19 11:13:54 INFO SparkContext: Job finished: count at <console>:13,
took 0.077148 s
14/08/19 11:14:03 INFO SparkContext: Job finished: count at <console>:13,
took 0.031572 s
```
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
