[ANNOUNCE] -humble gerrit/e2e-tests presentation

Marco Miller

unread,

Aug 7, 2020, 4:32:38 PM8/7/20

to Repo and Gerrit Discussion

Hey everyone,

Should you be interested in this topic in Subject, I'm now done preparing this presentation (also [1] below).

-At last, I should say. (It can be publicly viewed.)

Now, I'd like to go through it together with willing participants for an hour.

This would be scheduled for Wednesday August the 26th, at 10h30 am EST.

If you would like to participate, you can send me an email to marco....@gmail.com; double 'mm', here.

Please do NOT reply here for that specific matter.

-You're welcome to reply here though if otherwise.

This is so that I can add you to the list of meeting participants.

Note that your name/address will then be listed among the latter, in Google Calendar.

We will use Google Meet for that live discussion and demo.

If unable to join, no issue, as you can view the bulk of it online ([1]).

-I'm planning to add [1]'s link to [2] later on.

This doc ([1]) might move to some other, Gerrit standard location eventually; cf. [3].

[1] https://docs.google.com/presentation/d/1xZShuNKHmqeKAtfLzxwllQWze9P18i2nHbTzX_lQ9r4/edit?usp=sharing

[2] https://gerrit-documentation.storage.googleapis.com/Documentation/3.0.12/dev-e2e-tests.html

[3] https://bugs.chromium.org/p/gerrit/issues/detail?id=12208

Cheers then!

/Marco

Marco Miller

unread,

Aug 24, 2020, 3:40:00 PM8/24/20

to Repo and Gerrit Discussion

Reminder in case anyone missed this announcement below.

This is planned for after tomorrow, already; time flies.

/Marco

doug.r...@wandisco.com

unread,

Aug 26, 2020, 3:48:53 PM8/26/20

to Repo and Gerrit Discussion

Marco:

Very nice presentation today - Thank you!

Everyone:

I've got to ask how it is that this infrastructure does not enable

"first this then that" test steps? Having to guess/develop/iterate

until a "good value" for the step duration is found is highly

problematic. I question the scalability and robustness of any test

infrastructure critically based on step timeouts.

IMO would be better to go back and fix the plugin so that proper

gatling capabilities are enabled. Then the test cases developed

won't experience random breaks (due to, for instance, system

loading).

Cheers.

Doug

Marco Miller

unread,

Aug 26, 2020, 4:18:54 PM8/26/20

to Repo and Gerrit Discussion

On Wednesday, August 26, 2020 at 3:48:53 PM UTC-4 doug.r...@wandisco.com wrote:

Marco:

Very nice presentation today - Thank you!

And thank you for attending :)

Everyone:

I've got to ask how it is that this infrastructure does not enable
"first this then that" test steps? Having to guess/develop/iterate
until a "good value" for the step duration is found is highly
problematic. I question the scalability and robustness of any test
infrastructure critically based on step timeouts.

Good point and topic.

The framework I showed today does support first-then (sequential) test steps.

It is true though that such sequencing is limited to the currently known (expected) step allotted time.

IMO would be better to go back and fix the plugin so that proper
gatling capabilities are enabled. Then the test cases developed
won't experience random breaks (due to, for instance, system
loading).

This is a challenge indeed.

Not sure that it is a (gatling-git) plugin issue, though.

I showed framework resources that can help mitigate system-specific latencies.

-Improvements are welcome, yes.

Please do feel free to capture your precise ideas in Monorail [4] :)

[4] https://bugs.chromium.org/p/gerrit/issues/list?q=component%3Ae2e_tests&can=2&sort=priority