Thanks for the answer.
I found a workaround, not a fabulous one, but is works :)
pip3 install epdb pdbpp
# In the test
import epdb; epdb.serve()
# In accessing the issue
python3 -c 'import epdb; epdb.connect()'
I usually do this to debug apps running as system daemons.
epdb.serve() starts a server that allows connecting to the app
from the outside. Seems to be like a similar approach to what you
have in the docs for eclipse?
Anyways, if I do that at the beginning of the test and just continue
the test execution ( c ), it results in real time test output in the window
where I connected. :p
Be Well,
Alan
On Wed, Aug 15, 2018 at 11:30 AM, Lukáš Doktor <ldok...@redhat.com> wrote:
> Dne 15.8.2018 v 11:29 Lukáš Doktor napsal(a):
>> Dne 9.8.2018 v 21:44 Alan Martinovic napsal(a):
>>> Hey Cleber,
>>>
>>> yup, got the same behavior. Nice find :)
>>> These two behave the same:
>>>
>>> $ /tmp/my-shell-script.sh | while read f; do echo "$f"; done;
>>> $ /tmp/my-shell-script.sh
>>>
>>> And these two behave differently:
>>>
>>> $ /tmp/your-python-script.py # Prints line at a time
>>> $ /tmp/your-python-script.py | while read f; do echo "$f"; done; #
>>> Prints at once
>>>
>>
>> Hello Alan,
>>
>> python uses system default buffering, which for TTY devices is usually a
>> line and fully buffered on non-tty (eg. pipe). That is why you see different
>> behavior for `script.py | cat` vs. `script.py`.:
>> https://docs.python.org/2/library/functions.html#open
>>
>> If you still need immediate response, you need to open the stdout
>> unbuffered, or add `sys.stdout.flush()` after each write (note
>> sys.stdout.flush() is not guaranteed to flush the output immediately so some
>> delay might still happen, but in most cases it should suffice. Worst case
>> you'll also have to run `os.fsync(sys.stdout.fileno)`).
>>
>> This change should work well in your example. You might see another issue if
>> no new lines were to be generated, because Avocado does line buffering
>> internally. So something like:
>>
>> for AAA in `seq 1 10`; do echo -n $AAA; done
>>
>> Would wait until newline (or exit), but there is realistically nothing we
>> can do about this.
>
> I mean we can, we do support writing right-away, it's just that we decided
> not to do for performance reasons, so you'd have to convince us pretty
> strongly to disable line-buffering.
>
> Lukáš
>
>>
>> Regards,
>> Lukáš
>>
>>>
>>> Be Well,
>>> Alan
>>>
>>>
>>>
>>> On Thu, Aug 9, 2018 at 9:34 PM, Cleber Rosa <cr...@redhat.com> wrote:
>>>> Hi Alan,
>>>>
>>>> I can definitely reproduce your issue here. The interesting part is
>>>> that it (also) happens outside of Avocado. What I mean is:
>>>>
>>>> $ /tmp/my-shell-script.sh | while read f; do echo "$f"; done;
>>>>
>>>> Shows one line at a time. While:
>>>>
>>>> $ /tmp/your-python-script.py | while read f; do echo "$f"; done;
>>>>
>>>> Shows all of them at once. I need a bit more time to check the
>>>> differences here, and I'd appreciate if you could confirm that you get
>>>> the same behavior I'm describing here.
>>>>
>>>> But, if at all possible, Avocado should also try print the lines as
>>>> they're printed on both cases.
>>>>
>>>> Regards,
>>>> - Cleber.
>>>>
>>>> On 08/09/2018 03:14 PM, Alan Martinovic wrote:
>>>>> Hi Cleber,
>>>>> thank you for the response.
>>>>>
>>>>> Yes it's a SIMPLE test.
>>>>> In the normal scenario it's a test run on the host to check the behavior
>>>>> of an
>>>>> embedded Linux device via uart.
>>>>>
>>>>> But the behavior is the same also with this dummy test code:
>>>>>
>>>>> ---
>>>>>
>>>>> #!/usr/bin/python3
>>>>>
>>>>> import time
>>>>> import sys
>>>>>
>>>>> print("Reading through boot record")
>>>>> for i in range(10):
>>>>> time.sleep(2)
>>>>> print(i)
>>>>>
>>>>> sys.exit(1)
>>>>>
>>>>>
>>>>> ---
>>>>>
>>>>> I run it as
>>>>> $ avocado --show=test run dummy_test
>>>>>
>>>>> For this example it block until the test is completed and then prints
>>>>> the output. Will mark the part where it blocks in the output below.
>>>>>
>>>>>
>>>>> ---
>>>>>
>>>>>
>>>>> alan@alan-N551JM:~/workspace/platform_tests$ avocado --show=test run
>>>>> dummy_test
>>>>> found extension EntryPoint.parse('human = avocado.plugins.human:HumanJob')
>>>>> found extension EntryPoint.parse('jobscripts =
>>>>> avocado.plugins.jobscripts:JobScripts')
>>>>> found extension EntryPoint.parse('teststmpdir =
>>>>> avocado.plugins.teststmpdir:TestsTmpDir')
>>>>> found extension EntryPoint.parse('json_variants =
>>>>> avocado.plugins.json_variants:JsonVariants')
>>>>> File /etc/avocado/sysinfo/commands does not exist.
>>>>> File /etc/avocado/sysinfo/files does not exist.
>>>>> File /etc/avocado/sysinfo/profilers does not exist.
>>>>> Command line: /home/alan/.local/bin/avocado --show=test run dummy_test
>>>>>
>>>>> Avocado version: 63.0
>>>>>
>>>>> Config files read (in order):
>>>>> /home/alan/.local/lib/python3.6/site-packages/avocado/etc/avocado/avocado.conf
>>>>> /home/alan/.local/lib/python3.6/site-packages/avocado/etc/avocado/conf.d/resultsdb.conf
>>>>> /home/alan/.local/lib/python3.6/site-packages/avocado/etc/avocado/conf.d/jobscripts.conf
>>>>> /home/alan/.local/lib/python3.6/site-packages/avocado/etc/avocado/conf.d/gdb.conf
>>>>> /home/alan/.local/lib/python3.6/site-packages/avocado/etc/avocado/conf.d/result_upload.conf
>>>>> /home/alan/.config/avocado/avocado.conf
>>>>>
>>>>> Avocado config:
>>>>> Section.Key Value
>>>>> datadir.paths.base_dir /var/lib/avocado
>>>>> datadir.paths.test_dir /usr/share/doc/avocado/tests
>>>>> datadir.paths.data_dir /var/lib/avocado/data
>>>>> datadir.paths.logs_dir ~/avocado/job-results
>>>>> sysinfo.collect.enabled True
>>>>> sysinfo.collect.commands_timeout -1
>>>>> sysinfo.collect.installed_packages False
>>>>> sysinfo.collect.profiler False
>>>>> sysinfo.collect.locale C
>>>>> sysinfo.collect.per_test False
>>>>> sysinfo.collectibles.commands /etc/avocado/sysinfo/commands
>>>>> sysinfo.collectibles.files /etc/avocado/sysinfo/files
>>>>> sysinfo.collectibles.profilers /etc/avocado/sysinfo/profilers
>>>>> runner.output.colored True
>>>>> runner.output.utf8
>>>>> remoter.behavior.reject_unknown_hosts False
>>>>> remoter.behavior.disable_known_hosts False
>>>>> job.output.loglevel debug
>>>>> restclient.connection.hostname localhost
>>>>> restclient.connection.port 9405
>>>>> restclient.connection.username
>>>>> restclient.connection.password
>>>>> plugins.disable []
>>>>> plugins.skip_broken_plugin_notification []
>>>>> plugins.loaders ['file', '@DEFAULT']
>>>>> plugins.jobscripts.pre /etc/avocado/scripts/job/pre.d/
>>>>> plugins.jobscripts.post /etc/avocado/scripts/job/post.d/
>>>>> plugins.jobscripts.warn_non_existing_dir False
>>>>> plugins.jobscripts.warn_non_zero_status True
>>>>> gdb.paths.gdb /usr/bin/gdb
>>>>> gdb.paths.gdbserver /usr/bin/gdbserver
>>>>>
>>>>> Avocado Data Directories:
>>>>>
>>>>> base /home/alan/avocado
>>>>> tests /home/alan/.local/lib/python3.6/site-packages/examples/tests
>>>>> data /home/alan/avocado/data
>>>>> logs /home/alan/avocado/job-results/job-2018-08-09T21.10-a9168c7
>>>>>
>>>>> No variants available, using defaults only
>>>>>
>>>>> Variant : /
>>>>> Temporary dir: /var/tmp/avocado_u50x75ah/avocado_job_n6_sbv95
>>>>>
>>>>> Job ID: a9168c7da524d621b56947bea7651004f58ab551
>>>>>
>>>>> File /etc/avocado/sysinfo/commands does not exist.
>>>>> File /etc/avocado/sysinfo/files does not exist.
>>>>> File /etc/avocado/sysinfo/profilers does not exist.
>>>>> INIT 1-dummy_test
>>>>> PARAMS (key=timeout, path=*, default=None) => None
>>>>> Test metadata:
>>>>> filename: /home/alan/workspace/platform_tests/dummy_test
>>>>> teststmpdir: /var/tmp/avocado_44fpdrzb
>>>>> workdir: /var/tmp/avocado_u50x75ah/avocado_job_n6_sbv95/1-dummy_test
>>>>> START 1-dummy_test
>>>>> DATA (filename=output.expected) => NOT FOUND (data sources: variant, file)
>>>>> Running '/home/alan/workspace/platform_tests/dummy_test'
>>>>>
>>>>> ######### WAITING HERE.... #################
>>>>> ######### WAITING HERE.... #################
>>>>> ######### WAITING HERE.... #################
>>>>>
>>>>> ######### EVERYTHING PRINTED AT ONCE #################
>>>>>
>>>>> [stdout] Reading through boot record
>>>>> [stdout] 0
>>>>> [stdout] 1
>>>>> [stdout] 2
>>>>> [stdout] 3
>>>>> [stdout] 4
>>>>> [stdout] 5
>>>>> [stdout] 6
>>>>> [stdout] 7
>>>>> [stdout] 8
>>>>> [stdout] 9
>>>>> Command '/home/alan/workspace/platform_tests/dummy_test' finished with
>>>>> 1 after 20.04673981666565s
>>>>> Exit status: 1
>>>>> Duration: 20.04673981666565
>>>>>
>>>>> Reproduced traceback from:
>>>>> /home/alan/.local/lib/python3.6/site-packages/avocado/core/test.py:831
>>>>> Traceback (most recent call last):
>>>>> File
>>>>> "/home/alan/.local/lib/python3.6/site-packages/avocado/core/test.py",
>>>>> line 1123, in _execute_cmd
>>>>> env=test_params, encoding=defaults.ENCODING)
>>>>> File
>>>>> "/home/alan/.local/lib/python3.6/site-packages/avocado/utils/process.py",
>>>>> line 1271, in run
>>>>> raise CmdError(cmd, sp.result)
>>>>> avocado.utils.process.CmdError:
>>>>> ('/home/alan/workspace/platform_tests/dummy_test',
>>>>> <avocado.utils.process.CmdResult object at 0x7f53000ceba8>)
>>>>>
>>>>> During handling of the above exception, another exception occurred:
>>>>>
>>>>> Traceback (most recent call last):
>>>>> File
>>>>> "/home/alan/.local/lib/python3.6/site-packages/avocado/core/test.py",
>>>>> line 1179, in test
>>>>> self._execute_cmd()
>>>>> File
>>>>> "/home/alan/.local/lib/python3.6/site-packages/avocado/core/test.py",
>>>>> line 1128, in _execute_cmd
>>>>> raise exceptions.TestFail(details)
>>>>> avocado.core.exceptions.TestFail:
>>>>> ('/home/alan/workspace/platform_tests/dummy_test',
>>>>> <avocado.utils.process.CmdResult object at 0x7f53000ceba8>)
>>>>>
>>>>> Local variables:
>>>>> -> self <class 'avocado.core.test.SimpleTest'>: 1-dummy_test
>>>>> DATA (filename=output.expected) => NOT FOUND (data sources: variant, file)
>>>>> DATA (filename=stdout.expected) => NOT FOUND (data sources: variant, file)
>>>>> DATA (filename=stderr.expected) => NOT FOUND (data sources: variant, file)
>>>>> FAIL 1-dummy_test -> TestFail:
>>>>> ('/home/alan/workspace/platform_tests/dummy_test',
>>>>> <avocado.utils.process.CmdResult object at 0x7f53000ceba8>)
>>>>>
>>>>> Test results available in
>>>>> /home/alan/avocado/job-results/job-2018-08-09T21.10-a9168c7
>>>>>
>>>>> ---
>>>>>
>>>>> On Thu, Aug 9, 2018 at 8:05 PM, Cleber Rosa <cr...@redhat.com> wrote:
>>>>>>
>>>>>>
>>>>>> On 08/09/2018 01:40 PM, Alan Martinovic wrote:
>>>>>>> Hi,
>>>>>>>
>>>>>>> I'm running a long test as a standalone script which prints statuses to
>>>>>>> sdtout.
>>>>>>> I've noticed that avocado doesn't print the output out before the test
>>>>>>> finishes
>>>>>>> completely.
>>>>>>>
>>>>>>> As if it isn't "flushing lines" (probably not the correct term, but it
>>>>>>> explains the behavior :) )
>>>>>>>
>>>>>>> Is this a known issue/expected behavior or something to do with my
>>>>>>> environment?
>>>>>>>
>>>>>>>
>>>>>>> Be Well,
>>>>>>> Alan
>>>>>>>
>>>>>>
>>>>>> Hi Alan,
>>>>>>
>>>>>> IIUC, you're running what Avocado calls a SIMPLE test (an executable,
>>>>>> that returns zero/non-zero), right?
>>>>>>
>>>>>> I created a sample script to test the behavior you describe:
>>>>>>
>>>>>> ---
>>>>>>
>>>>>> #!/bin/sh
>>>>>> for i in $(seq 1 30); do
>>>>>> echo "Step $i";
>>>>>> sleep 1;
>>>>>> done;
>>>>>>
>>>>>> ---
>>>>>>
>>>>>> And ran:
>>>>>>
>>>>>> $ avocado --show=test run /tmp/script.sh
>>>>>>
>>>>>> I get one line per second printed, which is the expected behavior (line
>>>>>> buffered output).
>>>>>>
>>>>>> The same happens when the test output is sent to the log files. Can you
>>>>>> describe the behavior of your test? Does it print multiple lines? Or a
>>>>>> continuous stream of characters that are not separated by newlines?
>>>>>>
>>>>>> Regards!
>>>>>>
>>>>>> --
>>>>>> Cleber Rosa
>>>>>> [ Sr Software Engineer - Virtualization Team - Red Hat ]
>>>>>> [ Avocado Test Framework - avocado-framework.github.io ]
>>>>>> [ 7ABB 96EB 8B46 B94D 5E0F E9BB 657E 8D33 A5F2 09F3 ]
>>>>
>>>> --
>>>> Cleber Rosa
>>>> [ Sr Software Engineer - Virtualization Team - Red Hat ]
>>>> [ Avocado Test Framework - avocado-framework.github.io ]
>>>> [ 7ABB 96EB 8B46 B94D 5E0F E9BB 657E 8D33 A5F2 09F3 ]
>>>
>>
>>
>
>
