Github user roshannaik commented on the issue:
https://github.com/apache/storm/pull/2203
**ACKers=0, Workers=1**
Const->Id->Null | 1.x | v2-metrics | v2-no-cm
-- | -- | -- | --
elapsed (sec) | Throughput | Throughput | Throughput
60 | 553,881 | 557,223 | 294,525
120 | 644,973 | 741,814 | 391,532
180 | 355,677 | 419,571 | 298,619
240 | 126,078 | 157,305 | 289,647
300 | 79,578 | 85,359 | 317,680
360 | 50,575 | 48,993 | 201,650
420 | 32,841 | 25,075 | 136,033
480 | 17,437 | 13,695 | 56,827
540 | 9,572 | 6,871 | 67,272
TVL (rate=500k) | 1.x | v2-metrics | v2-no-cm
-- | -- | -- | --
elapsed (sec) | throughput | throughput | throughput
31 | 88,941.94 | 74,339.33 | 76,130.32
61 | 261,134.00 | 198,329.68 | 220,800.00
92 | 83,889.68 | 131,720.67 | 128,738.00
122 | 42,874.67 | 55,596.67 | 52,115.33
152 | 25,074.00 | 13,489.33 | 26,921.33
182 | 6,610.67 | -473,302.67 | 6,022.67
212 | 0 | 190,620.67 | -475,712.26
242 | -369,348.67 | 153,038.67 | 128,894.00
272 | 238,574.67 | 118,828.67 | 194,515.33
302 | 116,838.00 | 36,540.00 | 96,574.00
332 | 30,613.33 | 3,562.67 | 75,434.67
362 | 3,744.00 | -497,488.67 | 24,842.67
392 | -450,178.67 | 215,857.33 | 9,878.67
422 | 253,464.67 | 202,255.33 | 5,750.00
452 | 131,784.00 | 53,916.00 | 1,986.67
482 | 25,712.00 | 12,618.00 | -544,343.33
512 | 25,050.67 | 5,380.67 | 189,269.33
542 | 5,532.67 | -489,704.00 | 244,687.74
572 | 492 | 257,200.00 | 34,429.33
602 | -507,870.00 | 193,524.67 | 35,392.67
632 | 205,754.00 | 42,620.67 | 990.67
- In both topos and all binaries we see rapid degradation in throughput
eventually leading to worker dying and getting restarted. As mentioned before,
this is not a new concern for 1.x but being addressed in Storm 2.0.
- The workers for the v2-no-cm binaries ran a little longer than the others
before dying.
---
