cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Wilder Rodrigues <WRodrig...@schubergphilis.com>
Subject Re: [BLOCKER] - Guest VMs are not getting IPs as the DHCP port is not opened in VR
Date Wed, 16 Sep 2015 07:21:13 GMT
Hi all,

I did not say that the issues are not VR related, because they are. What I said was to not
try fixing it by calling the bash scripts form within the python files.

Those changes landed on master a few months ago, or more. The thing is that most of us rely
on the simulator and there was few people actually testing them. We tested a lot, as much
as we could at that time, but unfortunately the effort did not cover all the cases.

By the way, that’s not an excuse. We just lacked support from the community before, which
is actually being great now.

>From 6 blocker issues, 3 are VR related!

https://issues.apache.org/jira/secure/Dashboard.jspa?selectPageId=12326765

1 I already fixed, but cannot create a PR because master VPC router was broken after the 9th
September - which by the way, I have been reporting/looking into since Friday.

We have 52 PRs opened and few people reviewing them, I mean, really reviewing. Most of the
PRs haven’t been tested properly, which causes pain at the end.

Now, if we want 4.6 out, without issues, let’s work together and stop LGTM PRs that do not
contain:

1. Jira ticket
2. Proper description of what it fixes/improves
3. Unit and/or Integration tests
4. And was tested

After 4.6 our first priority should be to get a proper CI environment that not only run simulator,
but also KVM - at least.

my 0.20 cents

Cheers,
Wilder


On 16 Sep 2015, at 06:54, Rajani Karuturi <Rajani.Karuturi@citrix.com<mailto:Rajani.Karuturi@citrix.com>>
wrote:

I agree. Most of the blockers we see now are VR related.

~Rajani



On 16-Sep-2015, at 10:15 am, Raja Pullela <raja.pullela@citrix.com<mailto:raja.pullela@citrix.com>>
wrote:

I agree with jayapal, so far none of the BVTs for the hypervisors are at 100% except for the
simulator's.
Also, changes/refactoring some of the very core functionality such as VR should be thought
through and thoroughly tested before being pushed as the impact will be huge and stabilization
will take time.  At this point we are not sure what else is broken ???

Also, we should review/critic such changes to make sure we don't spend time on it later on...
Please note that Travis tests passing at 100% cannot be taken as the basis to think that the
changes are good.

my 2cents!
Raja
-----Original Message-----
From: Jayapal Reddy Uradi [mailto:jayapalreddy.uradi@citrix.com]
Sent: Wednesday, September 16, 2015 9:46 AM
To: dev@cloudstack.apache.org<mailto:dev@cloudstack.apache.org>
Subject: Re: [BLOCKER] - Guest VMs are not getting IPs as the DHCP port is not opened in VR

Hi Wilder,

CLOUDSTACK-8843<https://issues.apache.org/jira/browse/CLOUDSTACK-8843> This ticket is
actually NOT regression. It is not implemented/tested for the basic zone VR.
What I have observed in the VR refactor is that sanity testing of components the code affects
is not done.
Now we are spending time on fixing issues which were stable earlier.

Thanks,
Jayapal


On 15-Sep-2015, at 6:55 pm, Wilder Rodrigues <wrodrigues@schubergphilis.com<mailto:wrodrigues@schubergphilis.com><mailto:wrodrigues@schubergphilis.com>>
wrote:

Hi Jayapal,

I would not suggest to move back to the sceptres, but focus on fixing and for every fix test
it very well so we get confidence. If we start calling scripts com the python code it will
have a huge chance to become a real mess!

Let’s stick to it and fix it and once it’s stable we can discuss a refactor of the thing
from scratch - we already have some ideas.

The most important point is: for everyone touching Java/Python code, please test your changes
before creating PRs or LGTM other’s committers PRs.

Cheers,
Wilder


On 15 Sep 2015, at 14:52, Jayapal Reddy Uradi <jayapalreddy.uradi@citrix.com<mailto:jayapalreddy.uradi@citrix.com><mailto:jayapalreddy.uradi@citrix.com>>
wrote:

Wilder, That is true.

All these issues are because of the VR refactoring. I don’t know what are the unit test
cases run on VR refactor code.
VR (shell) scripts  are stabilized over years and now suddenly moved to new implementation.
What we have figured out is that this new implementation is done only partially. What I am
suggesting now to move ahead is to call the existing/old shell  scripts in new implementation
for configuring the VR to avoid spending time on unnecessary bugs.
With current implementation we may see many issues in production for some time to come.

Thanks,
Jayapal

On 15-Sep-2015, at 5:57 pm, Wilder Rodrigues <wrodrigues@schubergphilis.com<mailto:wrodrigues@schubergphilis.com><mailto:wrodrigues@schubergphilis.com>>
wrote:

I have the feeling that we are fixing 1 issue and created 2.

Please, make sure the routers tests are executed and green before creating the PR. I will
do the same.

Cheers,
Wilder


On 15 Sep 2015, at 14:23, Jayapal Reddy Uradi <jayapalreddy.uradi@citrix.com<mailto:jayapalreddy.uradi@citrix.com><mailto:jayapalreddy.uradi@citrix.com>>
wrote:

Assigned to myself.

Thanks,
Jayapal
On 15-Sep-2015, at 2:49 pm, Raja Pullela <raja.pullela@citrix.com<mailto:raja.pullela@citrix.com><mailto:raja.pullela@citrix.com>>
wrote:

Hi,

JIRA - ticket https://issues.apache.org/jira/browse/CLOUDSTACK-8843
Can someone please pick up this ticket ?  This is failing a bunch of the BVTs tests in Basiczone.

Thanks,
Raja








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