Open Stack

On Tue, Jan 26, 2016 at 07:32:13PM +0300, Ilya Shakhat wrote:
> 2016-01-26 17:56 GMT+03:00 Paul Belanger <pabelanger at redhat.com>:
> 
> 
> > > How about monitoring? There are 3 types of things to keep an eye on:
> > >
> > >    - front-end state - simple HTTP check that the server returns 200 and
> > >    some data
> > >    - data freshness - to check that the update was run during last couple
> > >    days - on s.com this is not automated, but it's something that my
> > >    colleagues care about
> > >    - errors in logs - rarely, but changes in module config in
> > default_data
> > >    may result in errors causing the whole project being non-updated --
> > >    currently monitored manually.
> > >
> > Do these checks exist today or something that needs to be added? If they
> > exist,
> > do you mind sharing the code? We do have cacti.openstack.org to provide
> > some
> > monitoring.  However, -infra only have manually checks at the moment.
> >
> 
> Only the first exist, others are something that would be nice to have. If
> keep them manual
> then someone will have to check things, at least from time to time.
> 
Okay, lets discussion next week with infra.

> 
> > > I still need to work on the drivelog stuff, I should have some time in
> > the
> > > > coming 2 weeks to finish that up.
> > > >
> > > I'm still hesitating whether driverlog should be a separate module, or it
> > > is just an additional config for Stackalytics. There are things that
> > > driverlog collects and Stackalytics does not (latest CI run on stable
> > > branches) but I'm not sure if anyone uses this data. So if you think that
> > > it would be easier to merge DriverLog into I would not disagree.
> > >
> > Might be worth bringing this up in an -infra meeting.  If the content is
> > static,
> > we can deal with that pretty easy. We can also add it to stackalytics.o.o,
> > however we are pretty tapped for memory ATM.
> >
> 
> Let's discuss this on the meeting next week (sorry, I cannot attend it
> today). I'll prepare
> information of memory consumption and other things,
> 
Ack'd
> 
> 
> > > > I'd like to formalize stackatlyics.o.o and see what is needed to move
> > it
> > > > into
> > > > production. This will allow us to move one step closer to 'Host
> > > > Stackalytics Service'[1]
> > > >
> > >
> > > Ok. So the remaining steps are:
> > >
> > >    - move the repo under openstack-infra
> > >    - change repo ACL to add Infra cores (I'm really-really looking for
> > >    other folks to review CRs)
> > >    - move default_data.json to project_config
> > >    - redirect s.com to s.o.o
> > >
> > Do we need to move the repo under openstack-infra, or just break some data
> > out?
> >
> 
> Well, we can keep the code at current location, I just copied the proposal
> from spec [1]
> 
> 
> > If we add some of the configuration into project-config, we'll get the
> > people
> > for code reviews.
> >
> > What is stopping us from doing the move today? I assume you need to update
> > some
> > code in stackalytics?
> >
> 
> Yes, there's reference to default_data.json from UI and defaults in config.
> Overall things to do:
>  - copy test case that validates default_data to project-config
>  - update links in stackalytics UI
>  - update docs
>  - update defaults in config
>  - send announcement
> 
Okay, let me talk with -infra and get this patch going.

> >
> >
> > > BTW, I know there is not much documentation on Stackalytics. Do I need to
> > > write some?
> > >
> > More docs help, I'd suggest adding it to
> > http://docs.openstack.org/infra/system-config/stackalytics.html
> >
> > > Best Regards,
> > > Ilya
> > >
> > > PS. Sorry for the late reply, the first email got lost during long
> > > holidays...
> > >
> > > 2016-01-21 18:30 GMT+03:00 Paul Belanger <pabelanger at redhat.com>:
> > >
> > > > On Tue, Jan 05, 2016 at 12:43:46PM -0500, Paul Belanger wrote:
> > > > > Hi Ilya,
> > > > >
> > > > > I wanted to see what else was needed from you to start working with
> > > > > stackalytics.openstack.org? Its been online for the last 6 weeks and
> > > > seems to be
> > > > > going strong.  Current maintenance is pretty limited but would be
> > > > interested in
> > > > > seeing what you need to do to the system to make it production ready.
> > > > >
> > > > > I still need to work on the drivelog stuff, I should have some time
> > in
> > > > the
> > > > > coming 2 weeks to finish that up.
> > > > >
> > > > > Bascially, I think we are at the point for
> > stackalytics.openstack.org
> > > > to handle
> > > > > some production traffic, but want to get your thoughts.
> > > > >
> > > > Ilya,
> > > >
> > > > Just following up on my original email, see if you had any questions or
> > > > concerns.
> > > >
> > > > I'd like to formalize stackatlyics.o.o and see what is needed to move
> > it
> > > > into
> > > > production. This will allow us to move one step closer to 'Host
> > > > Stackalytics Service'[1]
> > > >
> > > > [1]
> > > >
> > http://specs.openstack.org/openstack-infra/infra-specs/specs/stackalytics.html
> > > >
> >