Skip to content

bpo-36670: regrtest WindowsLoadTracker computes LOAD_FACTOR_1 #16556

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Oct 3, 2019
Merged

bpo-36670: regrtest WindowsLoadTracker computes LOAD_FACTOR_1 #16556

merged 1 commit into from
Oct 3, 2019

Conversation

vstinner
Copy link
Member

@vstinner vstinner commented Oct 3, 2019 •
edited by bedevere-bot
Loading

@vstinner vstinner requested a review from ammaraskar October 3, 2019 09:32
ammaraskar
ammaraskar approved these changes Oct 3, 2019
View reviewed changes
Copy link
Member

@ammaraskar ammaraskar left a comment

Choose a reason for hiding this comment

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

Looks good, and values seem consistent from some quick testing!

@vstinner
Copy link
Member Author

vstinner commented Oct 3, 2019

Looks good, and values seem consistent from some quick testing!

I compared the value computed load average to the moving average over the last minute: statistics.mean(values) where values are the 60 latest processor_queue_length.

The load average moves way more slowly and than the moving average. But it's part of the design of the load average, and Linux load average behaves the same.

My main worry is that Python load average always starts at 0.0 and so it takes at least one minute to reach the average. I modified the code to initialize the load to the first processor_queue_length. This is not perfect, but it helps to get an idea of the real load average in the first 60 seconds. After one minute, the value using the old and the new code should converge anyway.

--

I took this PR as an opportunity to push other enhancements and bugfixes.

@vstinner
Copy link
Member Author

vstinner commented Oct 3, 2019

I modified the code to initialize the load to the first processor_queue_length.

Example from the Win64 job of Azure Pipelines:

0:00:01 load avg: 9.00 [  5/419] test_list passed
...
0:00:30 load avg: 9.47 [ 35/419] test_multibytecodec passed
...
0:01:02 load avg: 8.36 [ 83/419] test_subprocess passed
...
0:02:00 load avg: 8.31 [186/419] test_faulthandler passed
...
0:03:05 load avg: 3.90 [226/419] test_cmd_line passed -- running: test_asyncio (48.1 sec), test_largefile (1 min 6 sec), test_io (32.3 sec)
...
...
...
0:04:44 load avg: 1.82 [248/419] test_urllib_response passed -- running: test_largefile (2 min 44 sec), test_io (2 min 10 sec)

Comparison with PR #16555:

0:00:00 load avg: 0.00 [  1/419] test_reprlib passed
...
0:00:30 load avg: 2.82 [ 49/419] test_source_encoding passed
...
0:01:04 load avg: 4.00 [102/419] test_ftplib passed -- running: test_asyncio (50 sec 188 ms)
...
0:01:30 load avg: 5.40 [144/419] test_types passed
...
...
...
0:11:25 load avg: 4.50 [415/419] test_distutils passed (1 min 19 sec)

@vstinner
bpo-36670: Enhance regrtest
d7e3840 
* Add log() method: add timestamp and load average prefixes
  to main messages.
* WindowsLoadTracker:

  * LOAD_FACTOR_1 is now computed using SAMPLING_INTERVAL
  * Initialize the load to the arithmetic mean of the first 5 values
    of the Processor Queue Length value (so over 5 seconds), rather
    than 0.0.
  * Handle BrokenPipeError and when typeperf exit.

* format_duration(1.5) now returns '1.5 sec', rather than
  '1 sec 500 ms'
@vstinner
Copy link
Member Author

vstinner commented Oct 3, 2019

I modified the code to initialize the load to the first processor_queue_length. This is not perfect, but it helps to get an idea of the real load average in the first 60 seconds. After one minute, the value using the old and the new code should converge anyway.

I made more tests and using a single value is really too raw. I modified my PR to use the arithmetic mean on the first 5 values instead. It helps to converge faster to the "real" load average.

Example of debug traces:

  • exp = computed load average
  • mean = arithmetic mean of the 60 latest values

The load average is initialized to 4.4 and it moves to 5.36 after 1 minute: 4.4 is a better estimation than 0.0 for the first value ;-)

exp: 4.4 -- mean: 4.40 -- last 10 values: 10, 9, 3, 0, 0
0:00:06 load avg: 4.40 running: test__osx_support (3.1 sec), test_heapq (3.1 sec)
exp: 4.3 -- mean: 3.67 -- last 10 values: 10, 9, 3, 0, 0, 0
0:00:07 load avg: 4.33 [  5/419] test__osx_support passed -- running: test_heapq (3.6 sec)
0:00:07 load avg: 4.33 [  6/419] test_heapq passed (1.6 sec)
0:00:07 load avg: 4.33 [  7/419] test_positional_only_arg passed
0:00:07 load avg: 4.33 [  8/419] test_crashers passed
exp: 4.4 -- mean: 4.00 -- last 10 values: 10, 9, 3, 0, 0, 0, 6
0:00:08 load avg: 4.35 running: test_grammar (1.0 sec), test_pathlib (1.0 sec)
exp: 4.3 -- mean: 3.62 -- last 10 values: 10, 9, 3, 0, 0, 0, 6, 1
0:00:09 load avg: 4.30 running: test_grammar (2.0 sec), test_pathlib (2.0 sec)
exp: 4.2 -- mean: 3.22 -- last 10 values: 10, 9, 3, 0, 0, 0, 6, 1, 0
0:00:10 load avg: 4.23 running: test_grammar (3.0 sec), test_pathlib (3.0 sec)
exp: 4.2 -- mean: 2.90 -- last 10 values: 10, 9, 3, 0, 0, 0, 6, 1, 0, 0
0:00:12 load avg: 4.16 running: test_grammar (4.0 sec), test_pathlib (4.0 sec)
exp: 4.1 -- mean: 2.64 -- last 10 values: 9, 3, 0, 0, 0, 6, 1, 0, 0, 0
0:00:12 load avg: 4.09 [  9/419] test_grammar passed -- running: test_pathlib (5.0 sec)
0:00:12 load avg: 4.09 [ 10/419] test_pathlib passed (4.5 sec)
exp: 4.2 -- mean: 3.33 -- last 10 values: 3, 0, 0, 0, 6, 1, 0, 0, 0, 11
0:00:13 load avg: 4.20 [ 11/419] test_dictcomps passed
0:00:13 load avg: 4.20 [ 12/419] test_filecmp passed
exp: 4.3 -- mean: 3.85 -- last 10 values: 0, 0, 0, 6, 1, 0, 0, 0, 11, 10
0:00:14 load avg: 4.30 running: test_argparse (1.0 sec), test_tcl (1.0 sec)
0:00:15 load avg: 4.30 [ 13/419] test_tcl passed -- running: test_argparse (1.2 sec)
exp: 4.3 -- mean: 3.93 -- last 10 values: 0, 0, 6, 1, 0, 0, 0, 11, 10, 5
0:00:15 load avg: 4.31 [ 14/419] test_codecmaps_kr passed -- running: test_argparse (1.8 sec)
0:00:16 load avg: 4.31 [ 15/419] test_imghdr passed -- running: test_argparse (2.5 sec)
exp: 4.4 -- mean: 4.20 -- last 10 values: 0, 6, 1, 0, 0, 0, 11, 10, 5, 8
0:00:16 load avg: 4.37 [ 16/419] test_netrc passed -- running: test_argparse (3.1 sec)
0:00:17 load avg: 4.37 [ 17/419] test_check_c_globals passed -- running: test_argparse (3.8 sec)
exp: 4.3 -- mean: 4.12 -- last 10 values: 6, 1, 0, 0, 0, 11, 10, 5, 8, 3
0:00:18 load avg: 4.35 [ 18/419] test_turtle passed -- running: test_argparse (4.5 sec)
exp: 4.4 -- mean: 4.24 -- last 10 values: 1, 0, 0, 0, 11, 10, 5, 8, 3, 6
0:00:18 load avg: 4.38 [ 19/419] test_argparse passed (4.6 sec)
0:00:18 load avg: 4.38 [ 20/419] test_format passed
0:00:19 load avg: 4.38 [ 21/419] test_kqueue skipped
test_kqueue skipped -- test works only on BSD
0:00:19 load avg: 4.38 [ 22/419] test_doctest2 passed
exp: 4.5 -- mean: 4.78 -- last 10 values: 0, 0, 0, 11, 10, 5, 8, 3, 6, 14
0:00:20 load avg: 4.54 [ 23/419] test_bigaddrspace passed
0:00:20 load avg: 4.54 [ 24/419] test_fork1 skipped
test_fork1 skipped -- object <module 'os' from 'C:\\vstinner\\python\\master\\\\lib\\os.py'> has no attribute 'fork'
0:00:20 load avg: 4.54 [ 25/419] test_defaultdict passed
exp: 4.6 -- mean: 4.84 -- last 10 values: 0, 0, 11, 10, 5, 8, 3, 6, 14, 6
0:00:21 load avg: 4.56 [ 26/419] test_genericclass passed -- running: test_idle (1.3 sec)
exp: 4.6 -- mean: 5.00 -- last 10 values: 0, 11, 10, 5, 8, 3, 6, 14, 6, 8
0:00:22 load avg: 4.62 running: test_idle (2.3 sec), test_set (1.0 sec)
0:00:22 load avg: 4.62 [ 27/419] test_idle passed (2.2 sec) -- running: test_set (1.4 sec)
exp: 4.6 -- mean: 5.05 -- last 10 values: 11, 10, 5, 8, 3, 6, 14, 6, 8, 6
0:00:23 load avg: 4.64 running: test_runpy (1.0 sec), test_set (2.4 sec)
exp: 4.7 -- mean: 5.23 -- last 10 values: 10, 5, 8, 3, 6, 14, 6, 8, 6, 9
0:00:24 load avg: 4.71 running: test_runpy (2.1 sec), test_set (3.5 sec)
exp: 4.7 -- mean: 5.22 -- last 10 values: 5, 8, 3, 6, 14, 6, 8, 6, 9, 5
0:00:26 load avg: 4.72 running: test_runpy (3.2 sec), test_set (4.6 sec)
exp: 4.7 -- mean: 5.25 -- last 10 values: 8, 3, 6, 14, 6, 8, 6, 9, 5, 6
0:00:26 load avg: 4.74 [ 28/419] test_set passed (5.1 sec) -- running: test_runpy (4.2 sec)
exp: 4.8 -- mean: 5.40 -- last 10 values: 3, 6, 14, 6, 8, 6, 9, 5, 6, 9
0:00:27 load avg: 4.81 [ 29/419] test_runpy passed (4.7 sec)
exp: 4.8 -- mean: 5.38 -- last 10 values: 6, 14, 6, 8, 6, 9, 5, 6, 9, 5
0:00:28 load avg: 4.81 running: test_csv (1.0 sec), test_tokenize (2.0 sec)
0:00:29 load avg: 4.81 [ 30/419] test_csv passed -- running: test_tokenize (2.2 sec)
exp: 4.8 -- mean: 5.41 -- last 10 values: 14, 6, 8, 6, 9, 5, 6, 9, 5, 6
0:00:30 load avg: 4.83 running: test_compileall (1.0 sec), test_tokenize (3.2 sec)
exp: 4.9 -- mean: 5.54 -- last 10 values: 6, 8, 6, 9, 5, 6, 9, 5, 6, 9
0:00:31 load avg: 4.90 running: test_compileall (2.1 sec), test_tokenize (4.3 sec)
exp: 4.9 -- mean: 5.59 -- last 10 values: 8, 6, 9, 5, 6, 9, 5, 6, 9, 7
0:00:32 load avg: 4.93 running: test_compileall (3.2 sec), test_tokenize (5.4 sec)
exp: 4.9 -- mean: 5.57 -- last 10 values: 6, 9, 5, 6, 9, 5, 6, 9, 7, 5
0:00:33 load avg: 4.94 [ 31/419] test_tokenize passed (5.5 sec) -- running: test_compileall (3.9 sec)
exp: 5.0 -- mean: 5.58 -- last 10 values: 9, 5, 6, 9, 5, 6, 9, 7, 5, 6
0:00:34 load avg: 4.95 running: test_compileall (4.9 sec), test_gettext (1.0 sec)
0:00:34 load avg: 4.95 [ 32/419] test_gettext passed (1.0 sec) -- running: test_compileall (5.5 sec)
exp: 5.0 -- mean: 5.66 -- last 10 values: 5, 6, 9, 5, 6, 9, 7, 5, 6, 8
0:00:35 load avg: 5.00 [ 33/419] test_c_locale_coercion passed -- running: test_compileall (6.4 sec)
exp: 5.0 -- mean: 5.67 -- last 10 values: 6, 9, 5, 6, 9, 7, 5, 6, 8, 6
0:00:36 load avg: 5.02 [ 34/419] test_with passed -- running: test_compileall (7.1 sec)
0:00:36 load avg: 5.02 [ 35/419] test_binhex passed -- running: test_compileall (7.7 sec)
exp: 5.1 -- mean: 5.71 -- last 10 values: 9, 5, 6, 9, 7, 5, 6, 8, 6, 7
0:00:37 load avg: 5.05 [ 36/419] test_generators passed -- running: test_compileall (8.5 sec)
exp: 5.3 -- mean: 6.06 -- last 10 values: 5, 6, 9, 7, 5, 6, 8, 6, 7, 18
0:00:38 load avg: 5.27 running: test_compileall (9.7 sec), test_smtpnet (1.2 sec)
0:00:39 load avg: 5.27 [ 37/419] test_smtpnet skipped (resource denied) -- running: test_compileall (9.9 sec)
test_smtpnet skipped -- Use of the 'network' resource not enabled
exp: 5.4 -- mean: 6.22 -- last 10 values: 6, 9, 7, 5, 6, 8, 6, 7, 18, 12
0:00:40 load avg: 5.38 running: test_compileall (10.9 sec), test_inspect (1.0 sec)
exp: 5.5 -- mean: 6.32 -- last 10 values: 9, 7, 5, 6, 8, 6, 7, 18, 12, 10
0:00:41 load avg: 5.45 running: test_compileall (12.1 sec), test_inspect (2.2 sec)
exp: 5.6 -- mean: 6.47 -- last 10 values: 7, 5, 6, 8, 6, 7, 18, 12, 10, 12
0:00:42 load avg: 5.56 running: test_compileall (13.3 sec), test_inspect (3.4 sec)
0:00:42 load avg: 5.56 [ 38/419] test_inspect passed (3.0 sec) -- running: test_compileall (13.7 sec)
exp: 5.6 -- mean: 6.54 -- last 10 values: 5, 6, 8, 6, 7, 18, 12, 10, 12, 9
0:00:43 load avg: 5.62 running: test_compileall (14.7 sec), test_xmlrpc (1.0 sec)
exp: 5.5 -- mean: 6.38 -- last 10 values: 6, 8, 6, 7, 18, 12, 10, 12, 9, 0
0:00:44 load avg: 5.53 running: test_compileall (15.7 sec), test_xmlrpc (2.0 sec)
exp: 5.4 -- mean: 6.22 -- last 10 values: 8, 6, 7, 18, 12, 10, 12, 9, 0, 0
0:00:45 load avg: 5.44 running: test_compileall (16.7 sec), test_xmlrpc (3.0 sec)
exp: 5.3 -- mean: 6.07 -- last 10 values: 6, 7, 18, 12, 10, 12, 9, 0, 0, 0
0:00:46 load avg: 5.35 running: test_compileall (17.7 sec), test_xmlrpc (4.0 sec)
exp: 5.3 -- mean: 5.93 -- last 10 values: 7, 18, 12, 10, 12, 9, 0, 0, 0, 0
0:00:47 load avg: 5.26 running: test_compileall (18.7 sec), test_xmlrpc (5.0 sec)
exp: 5.3 -- mean: 6.02 -- last 10 values: 18, 12, 10, 12, 9, 0, 0, 0, 0, 10
0:00:48 load avg: 5.34 running: test_compileall (19.8 sec), test_xmlrpc (6.1 sec)
exp: 5.2 -- mean: 5.89 -- last 10 values: 12, 10, 12, 9, 0, 0, 0, 0, 10, 0
0:00:49 load avg: 5.25 running: test_compileall (20.8 sec), test_xmlrpc (7.2 sec)
exp: 5.2 -- mean: 5.76 -- last 10 values: 10, 12, 9, 0, 0, 0, 0, 10, 0, 0
0:00:50 load avg: 5.16 running: test_compileall (21.8 sec), test_xmlrpc (8.2 sec)
exp: 5.1 -- mean: 5.64 -- last 10 values: 12, 9, 0, 0, 0, 0, 10, 0, 0, 0
0:00:51 load avg: 5.08 running: test_compileall (22.8 sec), test_xmlrpc (9.2 sec)
0:00:52 load avg: 5.08 [ 39/419] test_xmlrpc passed (9.2 sec) -- running: test_compileall (23.3 sec)
exp: 5.1 -- mean: 5.62 -- last 10 values: 9, 0, 0, 0, 0, 10, 0, 0, 0, 5
0:00:53 load avg: 5.07 running: test_compileall (24.4 sec), test_decimal (1.0 sec)
exp: 5.1 -- mean: 5.61 -- last 10 values: 0, 0, 0, 0, 10, 0, 0, 0, 5, 5
0:00:54 load avg: 5.07 running: test_compileall (25.5 sec), test_decimal (2.1 sec)
exp: 5.2 -- mean: 5.70 -- last 10 values: 0, 0, 0, 10, 0, 0, 0, 5, 5, 10
0:00:55 load avg: 5.15 running: test_compileall (26.6 sec), test_decimal (3.2 sec)
exp: 5.2 -- mean: 5.71 -- last 10 values: 0, 0, 10, 0, 0, 0, 5, 5, 10, 6
0:00:56 load avg: 5.17 running: test_compileall (27.7 sec), test_decimal (4.3 sec)
exp: 5.2 -- mean: 5.77 -- last 10 values: 0, 10, 0, 0, 0, 5, 5, 10, 6, 9
0:00:57 load avg: 5.23 running: test_compileall (28.8 sec), test_decimal (5.5 sec)
exp: 5.3 -- mean: 5.83 -- last 10 values: 10, 0, 0, 0, 5, 5, 10, 6, 9, 9
0:00:59 load avg: 5.29 running: test_compileall (29.9 sec), test_decimal (6.6 sec)
exp: 5.4 -- mean: 5.89 -- last 10 values: 0, 0, 0, 5, 5, 10, 6, 9, 9, 9
0:01:00 load avg: 5.36 running: test_compileall (31.0 sec), test_decimal (7.7 sec)
exp: 5.3 -- mean: 5.87 -- last 10 values: 0, 0, 5, 5, 10, 6, 9, 9, 9, 5
0:01:01 load avg: 5.35 running: test_compileall (32.0 sec), test_decimal (8.7 sec)
exp: 5.4 -- mean: 5.88 -- last 10 values: 0, 5, 5, 10, 6, 9, 9, 9, 5, 6
0:01:02 load avg: 5.36 running: test_compileall (33.1 sec), test_decimal (9.8 sec)
exp: 5.5 -- mean: 6.05 -- last 10 values: 5, 5, 10, 6, 9, 9, 9, 5, 6, 16
0:01:03 load avg: 5.54 running: test_compileall (34.2 sec), test_decimal (10.8 sec)
exp: 5.6 -- mean: 6.16 -- last 10 values: 5, 10, 6, 9, 9, 9, 5, 6, 16, 12
0:01:04 load avg: 5.64 running: test_compileall (35.2 sec), test_decimal (11.9 sec)
exp: 5.8 -- mean: 6.29 -- last 10 values: 10, 6, 9, 9, 9, 5, 6, 16, 12, 14
0:01:04 load avg: 5.78 [ 40/419] test_compileall passed (34.9 sec) -- running: test_decimal (12.1 sec)

@vstinner
Copy link
Member Author

vstinner commented Oct 3, 2019

(Comment on my PR #16555) @ammaraskar: "If you feel like 1 second is better, then the new constant would be 0.98347145382161748947374 https://www.wolframalpha.com/input/?i=1%2Fexp%281%2F60%29 "

Linux updates the load average every 5 seconds. But it's computed in the kernel. The problem here is that we start "from zero". I chose to compute the load average every second to converge faster to the "real" load average. The value of 1 second also helps my current implementation to compute the initial value. Getting samples every second and use the 5 first values to initial the load means that we can only seen the first load value after 5 seconds.

Getting samples every 5 seconds would mean that using my current implementation, we could only see the first load value after 25 seconds.

Or we could continue to initialize the load to 0.0, but it would mean that the first values before the first minute would be less accurate.

@vstinner
Copy link
Member Author

vstinner commented Oct 3, 2019

New logs with my latest PR, from Azure Pipelines.

The load was initialized to 8.37, whereas it reached 4.96 after 1 minute. It seems like the first value was a little bit "far" from the "real" load.

0:00:06 [ 16/419] test_asyncore passed
0:00:09 load avg: 8.37 [ 17/419] test_imp passed
...
0:00:30 load avg: 7.19 [ 36/419] test_ensurepip passed
...
0:01:01 load avg: 4.96 [ 47/419] test_pdb passed -- running: test_mmap (55.7 sec), test_mailbox (30.3 sec), test_asyncio (54.7 sec)
...
0:01:33 load avg: 3.80 [ 63/419] test_multiprocessing_main_handling passed -- running: test_mmap (1 min 28 sec), test_mailbox (1 min 3 sec)
...
0:02:00 load avg: 2.62 [ 83/419] test_telnetlib passed -- running: test_mmap (1 min 54 sec), test_mailbox (1 min 29 sec)
...
...
...
0:10:08 load avg: 3.98 [346/419] test_winsound passed -- running: test_largefile (1 min 46 sec), test_io (59.1 sec), test_zipfile (1 min 10 sec)

@vstinner
Copy link
Member Author

vstinner commented Oct 3, 2019

My previous comment was for Win64. Logs from Win32 job:

First value of 3.40 => 3.20 after 1 minute.

0:00:07 [ 14/419] test_pydoc passed
0:00:08 load avg: 3.40 [ 15/419] test_threadedtempfile passed
...
0:00:30 load avg: 2.62 [ 36/419] test_complex passed
...
0:01:07 load avg: 3.20 [ 64/419] test_multiprocessing_main_handling passed -- running: test_gettext (30.1 sec), test_largefile (1 min 3 sec)
...
0:01:30 load avg: 3.56 [104/419] test_list passed -- running: test_largefile (1 min 27 sec)
...
0:02:00 load avg: 3.50 [135/419] test_htmlparser passed -- running: test_largefile (1 min 57 sec)
...
...
...
0:11:40 load avg: 0.97 [419/419] test_io passed (3 min 26 sec)

@vstinner vstinner merged commit 098e256 into python:master Oct 3, 2019
@vstinner vstinner deleted the typeperf_load branch October 3, 2019 14:15
@vstinner vstinner mentioned this pull request Oct 3, 2019
Closed
vstinner added a commit that referenced this pull request Oct 3, 2019
@vstinner
[3.8] bpo-36670, regrtest: Fix WindowsLoadTracker() for partial line (G…
de3195c 
…H-16550) (GH-16560)

* bpo-36670, regrtest: Fix WindowsLoadTracker() for partial line (GH-16550)

WindowsLoadTracker.read_output() now uses a short buffer for
incomplete line.

(cherry picked from commit 3e04cd2)

* bpo-36670: Enhance regrtest WindowsLoadTracker (GH-16553)

The last line is now passed to the parser even if it does not end
with a newline, but only if it's a valid value.

(cherry picked from commit c65119d)

* bpo-36670: Enhance regrtest (GH-16556)

* Add log() method: add timestamp and load average prefixes
  to main messages.
* WindowsLoadTracker:

  * LOAD_FACTOR_1 is now computed using SAMPLING_INTERVAL
  * Initialize the load to the arithmetic mean of the first 5 values
    of the Processor Queue Length value (so over 5 seconds), rather
    than 0.0.
  * Handle BrokenPipeError and when typeperf exit.

* format_duration(1.5) now returns '1.5 sec', rather than
  '1 sec 500 ms'

(cherry picked from commit 098e256)
@miss-islington miss-islington mentioned this pull request Oct 3, 2019
Merged
miss-islington pushed a commit to miss-islington/cpython that referenced this pull request Oct 3, 2019
@vstinner @miss-islington
[3.8] bpo-36670, regrtest: Fix WindowsLoadTracker() for partial line (p…
979a4ae 
…ythonGH-16550) (pythonGH-16560)

* bpo-36670, regrtest: Fix WindowsLoadTracker() for partial line (pythonGH-16550)

WindowsLoadTracker.read_output() now uses a short buffer for
incomplete line.

(cherry picked from commit 3e04cd2)

* bpo-36670: Enhance regrtest WindowsLoadTracker (pythonGH-16553)

The last line is now passed to the parser even if it does not end
with a newline, but only if it's a valid value.

(cherry picked from commit c65119d)

* bpo-36670: Enhance regrtest (pythonGH-16556)

* Add log() method: add timestamp and load average prefixes
  to main messages.
* WindowsLoadTracker:

  * LOAD_FACTOR_1 is now computed using SAMPLING_INTERVAL
  * Initialize the load to the arithmetic mean of the first 5 values
    of the Processor Queue Length value (so over 5 seconds), rather
    than 0.0.
  * Handle BrokenPipeError and when typeperf exit.

* format_duration(1.5) now returns '1.5 sec', rather than
  '1 sec 500 ms'

(cherry picked from commit 098e256)
(cherry picked from commit de3195c)

Co-authored-by: Victor Stinner <vstinner@python.org>
miss-islington added a commit that referenced this pull request Oct 3, 2019
@miss-islington @vstinner
[3.8] bpo-36670, regrtest: Fix WindowsLoadTracker() for partial line (G…
f9016e5 
…H-16550) (GH-16560)

* bpo-36670, regrtest: Fix WindowsLoadTracker() for partial line (GH-16550)

WindowsLoadTracker.read_output() now uses a short buffer for
incomplete line.

(cherry picked from commit 3e04cd2)

* bpo-36670: Enhance regrtest WindowsLoadTracker (GH-16553)

The last line is now passed to the parser even if it does not end
with a newline, but only if it's a valid value.

(cherry picked from commit c65119d)

* bpo-36670: Enhance regrtest (GH-16556)

* Add log() method: add timestamp and load average prefixes
  to main messages.
* WindowsLoadTracker:

  * LOAD_FACTOR_1 is now computed using SAMPLING_INTERVAL
  * Initialize the load to the arithmetic mean of the first 5 values
    of the Processor Queue Length value (so over 5 seconds), rather
    than 0.0.
  * Handle BrokenPipeError and when typeperf exit.

* format_duration(1.5) now returns '1.5 sec', rather than
  '1 sec 500 ms'

(cherry picked from commit 098e256)
(cherry picked from commit de3195c)

Co-authored-by: Victor Stinner <vstinner@python.org>
jacobneiltaylor pushed a commit to jacobneiltaylor/cpython that referenced this pull request Dec 5, 2019
@vstinner
bpo-36670: Enhance regrtest (pythonGH-16556)
3c5f948 
* Add log() method: add timestamp and load average prefixes
  to main messages.
* WindowsLoadTracker:

  * LOAD_FACTOR_1 is now computed using SAMPLING_INTERVAL
  * Initialize the load to the arithmetic mean of the first 5 values
    of the Processor Queue Length value (so over 5 seconds), rather
    than 0.0.
  * Handle BrokenPipeError and when typeperf exit.

* format_duration(1.5) now returns '1.5 sec', rather than
  '1 sec 500 ms'
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@ammaraskar ammaraskar ammaraskar approved these changes

Assignees
No one assigned
Labels
skip news
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@vstinner @ammaraskar @the-knights-who-say-ni @bedevere-bot