cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bharat Kumar <bharat.ku...@citrix.com>
Subject Re: Blameless post mortem
Date Mon, 28 Sep 2015 12:13:37 GMT
Hi Sebastien,

You are confused, we are talking about  persistent VR config changes. below is the pr related
to it.
https://github.com/apache/cloudstack/pull/118

If you look at it you will notice that there are more than 250 commits and only a few tests
that were run.

Regards,
Bharat.

On 28-Sep-2015, at 5:24 pm, Bharat Kumar <bharat.kumar@citrix.com<mailto:bharat.kumar@citrix.com>>
wrote:

Hi Remi,

Whatever  ever we think  we have discovered are all well known best practices while developing
code in community.
I agree that tests need to be run on a new PR,  but i wonder why was this ignored when merging
the VR refactor code. Perhaps we will uncover some more issues if we investigate this. I believe
one of the reasons for this is the complexity and incomplete nature of the vr refactor change
and failing to identify the areas which got effected. If we had a good documentation i think
we cloud have understood the areas that were getting
impacted early on, areas like the vpn ,  iptables, isolated networks rvr   etc  and run the
relevant tests. The documentation will also help us focus on these areas while reviewing 
and fixing subsequent issues. Currently no one knows the areas that got effected
due to the vr refactor change, we are seeing issues all over the code.  I think this is a
bigger problem than what we have discussed so far.

I think presently we should stop fixing all the vr refactoring  bugs until we come up with
a  proper document describing the VR refactoring  changes.

I am not suggesting that we should revert the vr refactor code, I am willing to work on this
and fix the issues,  I am only asking if we can get some documentation.


Regards,
Bharat.

On 28-Sep-2015, at 4:59 pm, Sebastien Goasguen <runseb@gmail.com<mailto:runseb@gmail.com>>
wrote:


On Sep 28, 2015, at 1:14 PM, Remi Bergsma <RBergsma@schubergphilis.com<mailto:RBergsma@schubergphilis.com>>
wrote:

Hi Bharat,


There is only one way to prove a feature works: with tests. That’s why I say actually _running_
the tests we have today on any new PR, is the most important thing. Having no documentation
is a problem, I agree, but it is not more important IMHO. If we had the documentation, we
still would have issues if nobody runs the tests and verifies pull requests. Documentation
that is perfect does not automatically lead to perfect implementation. So we need tests to
verify.

If we don’t agree that is also fine. We need to do both anyway and I think we do agree on
that.


Also we need to move forward. We should have a live chat once 4.6 is out to discuss all issues/problems
and iron out the process.

But reverting the VR refactor is not going to happen. There was ample discussions on the PR
when it was submitted, there was time to review and raise concerns at that time. It went through
quite a few reviews, tests etc…Maybe the documentation is not good, but the time to raise
this concern I am afraid was six months ago. We can learn from it, but we are not going to
revert it, this would not go cleanly as David mentioned.

So let’s get back to blockers for 4.6, are there still VR related issues with master ?




Regards,
Remi






On 28/09/15 12:15, "Bharat Kumar" <bharat.kumar@citrix.com<mailto:bharat.kumar@citrix.com>>
wrote:

Hi Remi,

i do not agree with “There is no bigger problem”  part of your reply. so I had to repeat
myself to make it more clear, Not because i am not aware of what this thread is supposed to
do.

Regards,
Bharat.

On 28-Sep-2015, at 2:51 pm, Remi Bergsma <RBergsma@schubergphilis.com<mailto:RBergsma@schubergphilis.com>>
wrote:

Hi Bharat,

I understand your frustrations but we already agreed on this so no need to repeat. This thread
is supposed to list some improvements and learn from it. Your point has been taken so let’s
move on.

We need documentation first, then do a change after which all tests should pass. Even better
is we write (missing) tests before changing stuff so you know they pass before and after the
fact.

When doing reviews, feel free to ask for design documentation if you feel it is needed.

Regards, Remi



On 28/09/15 11:02, "Bharat Kumar" <bharat.kumar@citrix.com<mailto:bharat.kumar@citrix.com>>
wrote:

Hi Remi,

I never intended to say that we should not run tests, but even before tests we should have
proper documentation. My concern was if a major change is being introduced it should be properly
documented. All the issues which we are trying to fix are majorly due to VR refactor. If there
was a proper documentation for this we could
have fixed this in a better way.  Even to add tests we need to understand how a particular
thing works and what data dose it expect. I think while fixing the python based code changes
this is where most of the people are facing issues. A proper documentation will help in understanding
these in a better way.

Thanks,
Bharat.

On 28-Sep-2015, at 1:57 pm, Remi Bergsma <RBergsma@schubergphilis.com<mailto:RBergsma@schubergphilis.com>>
wrote:

Hi Bharat,

There is no bigger problem. We should always run the tests and if we find a case that isn’t
currently covered by the tests we should simply add tests for it. There’s no way we’ll
get a stable master without them. The fact that they may not cover everything, is no reason
to not rely on them. If a feature is not important enough to write a test for, then the feature
is probably not important anyway. And if it is, then add a test :-)

I do agree on the design documentation requirement for any (major?) change. I found some design
documentations on the subject you mention, but it should have been more detailed.

Regards,
Remi






On 28/09/15 09:58, "Bharat Kumar" <bharat.kumar@citrix.com<mailto:bharat.kumar@citrix.com>>
wrote:

Hi Remi,

Thank you for the Blame less postmortem.

I think there is a bigger problem here than just the review process and running tests. Even
if we run the tests we cannot be sure that every thing will work as intended. The tests will
only give some level of confidence. The tests may not cover all the use cases.

I think the problem here is that the way major changes to the code base are dealt with. For
example,  VR refactoring was done without discussing the design implications and the amount
of changes it would bring in. I could not find any design document. The vr refactor changed
a lot of code and the way VR used to work and in my opinion it was incomplete-vpn, isolated
networks, basic networks, iptable rules and rvr in isolated case etc were not implemented.
Most of us are still in the process of understanding this. Even before reaching this state
we had to spend a lot of time fixing issues mentioned in the thread [Blocker/Critical] VR
related Issues.

When a change of this magnitude is being made, we should call out all the changes and document
them properly. This will help people to create better fixes. Currently when we attempt to
fix the isolated vr case it is effecting the vpc and vice versa. for example pr 738 fixed
it for vpc networks but broke it for isolated case. I believe it is not too late to at least
start documenting the changes now.

Thanks,
Bharat.

On 28-Sep-2015, at 10:52 am, Sanjeev N <sanjeev@apache.org<mailto:sanjeev@apache.org>>
wrote:

I have a concern here. Some of us are actively involved in reviewing the
PRs related to marvin tests(Enhancing existing tests/Adding new tests). If
we have to test a PR it requires an environment to be created with actual
resources and this is going to take lot of time. Some of the tests can run
on simulator but most of the tests require real hardware to test. PR
submitter is already testing and submitting the test results along with the
PR. So is it require to test these PRs by reviewers?

On Sat, Sep 26, 2015 at 1:49 PM, sebgoa <runseb@gmail.com<mailto:runseb@gmail.com>>
wrote:

Remi, thanks for the detailed post-mortem, it's a good read and great
learning.
I hope everyone reads it.

The one thing to emphasize is that we now have a very visible way to get
code into master, we have folks investing time to provide review (great),
we need the submitters to make due diligence and answer all comments in the
reviews.

In another project i work on, nothing can be added to the code without
unit tests. I think we could go down the route of asking for new
integration tests and unit tests for anything. If not, the PR does not get
merged. But let's digest your post-mortem and we can discuss after 4.6.0.

I see that you reverted one commit that was not made by you, that's great.

Let's focus on the blockers now, everything else can wait.

The big bonus of doing what we are doing is that once 4.6.0 is out, we can
merge PRs again (assuming they are properly rebased and tested) and we can
release 4.6.1 really quickly after.

-sebastien

On Sep 25, 2015, at 9:51 PM, Remi Bergsma <RBergsma@schubergphilis.com<mailto:RBergsma@schubergphilis.com>>
wrote:

Hi all,

This mail is intended to be blameless. We need to learn something from
it. That's why I left out who exactly did what because it’s not relevant.
There are multiple examples but it's about the why. Let's learn from this
without blaming anyone.

We know we need automated testing. We have integration tests, but we are
unable to run all of them on any Pull Request we receive. If we would have
that in place, it'd be much easier to spot errors, regression and so on.
It'd also be more rewarding to write more tests.

Unfortunately we're not there yet. So, we need to do something else
instead until we get there. If we do nothing, we know we have many issues
because a master that breaks on a regular basis is the most frustrating
things. We said we'd use Pull Requests with at least two humans to review
and give their OK for a Pull Request. In the form of LGTM: Looks Good To
Me. Ok, so the LGTMs are there because we have no automated testing. Keep
that in mind. You are supposed to replace automated testing until it's
there.

Since we do this, master got a lot more stable. But every now and then
we still have issues. Let's look at how we do manual reviews. Again, this
is not to blame anyone. It's to open our eyes and make us realise what
we're doing and what results we get out of that.


Example Pull Request #784:
Title: CLOUDSTACK-8799 fixed the default routes

That's nice, it has a Jira id and a short description (as it should be).

The first person comes along and makes a comment:
"There was also an issue with VPC VRs" ... "Have you seen this issue?
Does your change affects the VPC VR (single/redundant)?"

Actually a good question. Unfortunaly there comes no answer. After a
reminder, it was promised to do tests against VPC networks. Great!

The Jenkins builds both succeed and also Travis is green. But how much
value does this have? They have the impression to do automated testing, and
although you could argue they do, it's far from complete. If it breaks, you
know you have an issue. But it doesn’t work the other way around.

Back to our example PR. In the mean time, another commit gets pushed to
it: "CLOUDSTACK-8799 fixed for vpc networks." But if you look at the Jira
issue, you see it is about redundant virtual routers. The non-VPC ones. So
this is vague at best. But a reviewer gives a LGTM because the person could
create a VPC. That doesn't have anything to do with the problem being fixed
in this PR nor with the comments made earlier. But, at least the person
said what he did and we should all do that. What nobody knew back then, was
that this broke the default route on VPCs.

Then something strange happens: the two commits from the PR end up on
master as direct commits. With just one LGTM and no verification from the
person commenting about the linked issue. This happened on Friday September
11th.

That day 21 commits came in, from 7 Pull Request and unfortunately also
from some direct commits. We noticed the direct commits and notified the
list (http://cloudstack.markmail.org/message/srmszloyipkxml36). As a lot
came in at the same time, it was decided not to revert them. Looking back,
we should have done it.

>From this point on, VPCs were broken as they wouldn't get a default
route. So, no public internet access from VMs in VPC tiers, no VPNs
working, etc. This was mentioned to the list on Thursday September 15th,
after some chats and debugging going on over the weekend (
http://cloudstack.markmail.org/message/73ulpu4p75ex24tc)

Here we are, master is broken functionality wise and new Pull Requests
come in to fix blockers. But we cannot ever test their proper working,
because VPCs are broken in master and so also in the PRs branched off of
it. With or without change in the PR.

It starts to escalate as the days go by.

I’ll leave out the bit on how this frustrated people. Although it’s good
to know we do not want to be in this situation.

Eventually Wilder and I spent an evening and a day working on a branch
where we loaded 7 PRs on top of each other (all VR related) only to find
the VPC is still broken. It allowed us to zoom in and find the default
route was missing again. We said it worked 3 weeks before, because the same
tests that succeeded then, now were broken. We had already fixed this in PR
#738 on August 25 so were sure about it.

After some digging we could trace it back to Pull Request #784. Imagine
the feeling seeing your own comment there mentioning the previous issue on
the default gateways. Fair to say our human review process clearly failed
here. Many many hours were spent on this problem over the past two weeks.
Could we have prevented this from happening? I think so, yes.


This example clearly shows why:

- we should use Pull Requests
It made the change visible: Great!

- we do reviews and ask for feedback
We got feedback and questions: Also great!

- we should always respond to feedback and verify it is resolved, before
merging
We need to improve here. Even with two reviewers that say LGTM, we
should still address any feedback before merging.

- we should have two humans doing a review
We need to improve here as well. Not one reviewer, we need two. Really.

- we need to document why we say LGTM.
Another improvement. It’s nice to say LGTM, but a review of only 4
characters and nothing more is useless. We need to know what was tested and
how. Test results, screen shots or anything that shows what's been
verified. If you only reviewed the code, also fine but at least say that.
Then the next reviewer should do another type of review to get the comlete
picture. Remember you're replacing automated testing!

- we should always merge Pull Requests
We made it easy, merging is the de facto standard, and it has even more
benefits. You can trace commits back to their Pull Request (and find all
comments and discussion there: saves time, trust me). It also allows for
easier reverting of a Pull Request. We’ll see even more benefits once 4.7
is there. Although the intentions to merge the Pull Request were there, it
still didn't happen. We should always check before we push. As a committer
we just need to be sure.

- we need automated testing!
The sooner the better. It’s all about the missing automated testing.
After 4.6, we all need to focus on this. Saves a lot of time. And
frustrations.



We're doing final testing on PR #887 and will merge it soon. From that
point on we can look into new issues. Be aware that any PR out there that
was created after September 10 needs to be rebased with current master
(when #887 is merged). Without that, no serious testing can be done.

Let's be careful what to land on master. I'll only be merging Pull
Requests that have had proper reviews with information on what was tested.
At least one reviewer needs to actually verify it works (and show the rest
of us). We simply cannot assume it will work.

If we do this, I think we can start resolving the remaining blockers
one-by-one and go into the first RC round. Please help out where you can so
we can make this a success together. Thanks!

Looking forward to the day we have our automated testing in place ;-)

Regards,
Remi










Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message