On 01/06/2014 07:04 PM, Joe Gordon wrote:
Overall this looks really good, and very spot on.
On Thu, Jan 2, 2014 at 6:29 PM, Sean Dague <s...@dague.net
<mailto:s...@dague.net>> wrote:
A lot of elastic recheck this fall has been based on the ad hoc
needs of the moment, in between diving down into the race bugs that
were uncovered by it. This week away from it all helped provide a
little perspective on what I think we need to do to call it *done*
(i.e. something akin to a 1.0 even though we are CDing it).
Here is my current thinking on the next major things that should
happen. Opinions welcomed.
(These are roughly in implementation order based on urgency)
= Split of web UI =
The elastic recheck page is becoming a mismash of what was needed at
the time. I think what we really have emerging is:
* Overall Gate Health
* Known (to ER) Bugs
* Unknown (to ER) Bugs - more below
I think the landing page should be Know Bugs, as that's where we
want both bug hunters to go to prioritize things, as well as where
people looking for known bugs should start.
I think the overall Gate Health graphs should move to the zuul
status page. Possibly as part of the collection of graphs at the bottom.
We should have a secondary page (maybe column?) of the
un-fingerprinted recheck bugs, largely to use as candidates for
fingerprinting. This will let us eventually take over /recheck.
I think it would be cool to collect the list of unclassified failures
(not by recheck bug), so we can see how many (and what percentage) need
to be classified. This isn't gate health but more of e-r health or
something like that.
Agreed. I've got the percentage in check_success today, but I agree that
every gate job that fails that we don't have a fingerprint should be
listed somewhere we can work through them.
= Data Analysis / Graphs =
I spent a bunch of time playing with pandas over break
(http://dague.net/2013/12/30/__ipython-notebook-experiments/
<http://dague.net/2013/12/30/ipython-notebook-experiments/>)__, it's
kind of awesome. It also made me rethink our approach to handling
the data.
I think the rolling average approach we were taking is more precise
than accurate. As these are statistical events they really need
error bars. Because when we have a quiet night, and 1 job fails at
6am in the morning, the 100% failure rate it reflects in grenade
needs to be quantified that it was 1 of 1, not 50 of 50.
So my feeling is we should move away from the point graphs we have,
and present these as weekly and daily failure rates (with graphs and
error bars). And slice those per job. My suggestion is that we do
the actual visualization with matplotlib because it's super easy to
output that from pandas data sets.
The one thing that the current graph does, that weekly and daily failure
rates don't show, is a sudden spike in one of the lines. If you stare
at the current graphs for long enough and can read through the noise,
you can see when the gate collectively crashes or if just the neutron
related gates start failing. So I think one more graph is needed.
The point of the visualizations is to make sense to people that don't
understand all the data, especially core members of various teams that
are trying to figure out "if I attack 1 bug right now, what's the
biggest bang for my buck."
Basically we'll be mining Elastic Search -> Pandas TimeSeries ->
transforms and analysis -> output tables and graphs. This is
different enough from our current jquery graphing that I want to get
ACKs before doing a bunch of work here and finding out people don't
like it in reviews.
Also in this process upgrade the metadata that we provide for each
of those bugs so it's a little more clear what you are looking at.
For example?
We should always be listing the bug title, not just the number. We
should also list what projects it's filed against. I've stared at this
bugs as much as anyone, and I still need to click through the top 4 to
figure out which one is the ssh bug. :)
= Take over of /recheck =
There is still a bunch of useful data coming in on "recheck bug
####" data which hasn't been curated into ER queries. I think the
right thing to do is treat these as a work queue of bugs we should
be building patterns out of (or completely invalidating). I've got a
preliminary gerrit bulk query piece of code that does this, which
would remove the need of the daemon the way that's currently
happening. The gerrit queries are a little long right now, but I
think if we are only doing this on hourly cron, the additional load
will be negligible.
This would get us into a single view, which I think would be more
informative than the one we currently have.
treating /recheck as a work queue sounds great, but this needs a bit
more fleshing out I think.
I imagine the workflow as something like this:
* State 1: Path author files bug saying 'gate broke, I didn't do it and
don't know why it broke'.
* State 2: Someone investigates the bug and determines if bug is valid
and if its a duplicate or not. root cause still isn't known.
* State 3: Someone writes a fingerprint for this bug and commits it to
elastic-recheck.
Assuming we agree on this general workflow, it would be nice if /recheck
distinguished between bugs in states 1 and 2, and there is no need to
list bugs in state 3 as e-r bot will automatically tell a developer when
he hits it.
Sure, that means policy on something in the bugs that can distinguish
between. I assume LP states.
State 1 = new & invalid?
State 2 = confirmed / triaged?
I think we can call that post 1.0 though, as we'll be adding details
beyond anything we have today.
= Categorize all the jobs =
We need a bit of refactoring to let us comment on all the jobs (not
just tempest ones). Basically we assumed pep8 and docs don't fail in
the gate at the beginning. Turns out they do, and are good
indicators of infra / external factor bugs. They are a part of the
story so we should put them in.
Don't forget grenade
Yep. That's part of all. :) I was just calling out the others as
something not originally on the list.
= Multi Line Fingerprints =
We've definitely found bugs where we never had a really satisfying
single line match, but we had some great matches if we could do
multi line.
We could do that in ER, however it will mean giving up logstash as
our UI, because those queries can't be done in logstash. So in order
to do this we'll really need to implement some tools - cli minimum,
which will let us easily test a bug. A custom web UI might be in
order as well, though that's going to be it's own chunk of work,
that we'll need more volunteers for.
This would put us in a place where we should have all the
infrastructure to track 90% of the race conditions, and talk about
them in certainty as 1%, 5%, 0.1% bugs.
Horrah. multi line matches are two separate elasticSearch queries, where
you match build_uuids. So to get the set of all hits of a multi line
fingerprint you find the intersection between line_1 and line_2 where
the key is build_uuid
Yes. The biggest issue is tooling for making it easy for people to test
their queries. It's pretty unfriendly to tell people to do manual
correlation in ES.
-Sean
--
Sean Dague
Samsung Research America
s...@dague.net / sean.da...@samsung.com
http://dague.net
_______________________________________________
OpenStack-dev mailing list
OpenStack-dev@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
