Have the CQ commit when the tree is "throttled" - Chromium-dev

The old Google Groups will be going away soon, but your browser is incompatible with the new version.
Message from discussion Have the CQ commit when the tree is "throttled"
Received-SPF: pass (google.com: domain of v...@google.com designates 209.85.210.175 as permitted sender) client-ip=209.85.210.175;
MIME-Version: 1.0
Sender: v...@google.com
In-Reply-To: <CAEoffTAsQM6f=Vd8BHByWdFFZoPAJhMthzP4Lb8=Mbj7FDH...@mail.gmail.com>
References: <CANAQWOVSPzNtzJWBKW12-uhLaAYoJafxZfEL52LOpR4hX5n...@mail.gmail.com>
	<CAEoffTBrisPycKEKrmpaTS5zR3NU3np4xOUu0hLyuVVmjhj...@mail.gmail.com>
	<CAHtyhaTH=n7k_+aMVxdhwsD-DG7F1jzdR8GhLVJW2TtO=G0...@mail.gmail.com>
	<CANAQWOXNm5v4bd=QZTX7sOC1KWDsA5+TxG40oY0jj7gpu_Z...@mail.gmail.com>
	<CAEoffTAsQM6f=Vd8BHByWdFFZoPAJhMthzP4Lb8=Mbj7FDH...@mail.gmail.com>
Date: Thu, 15 Nov 2012 23:55:22 -0800
Message-ID: <CABqr5co1-2KSB6WE=gsuRfaSTRFRu8hDXi-+zb0DBhrsqM=...@mail.gmail.com>
Subject: Re: [chromium-dev] Have the CQ commit when the tree is "throttled"
From: Viet-Trung Luu <viettrung...@chromium.org>
To: dpra...@google.com
Cc: Marc-Antoine Ruel <mar...@google.com>, Dana Jansens <dan...@chromium.org>, maruel+leg...@google.com, 
	chromium-dev <chromium-...@chromium.org>
Content-Type: multipart/alternative; boundary=90e6ba6138d0f658f604ce98167a

--90e6ba6138d0f658f604ce98167a
Content-Type: text/plain; charset=UTF-8

+1 to Dirk's concerns.

As sheriff, I'd throttle to let some bots catch up. I might entertain
commit requests to a throttled tree, but I'd vet them (e.g., if I'm waiting
for Win bots to catch up, I might let a Mac-only change in).

I don't see why the CQ should get more privileges with respect to a
throttled tree. A manual commit may/should also come with a complete set of
try jobs, and moreover the committer is more likely to be online and
responsive!

That we throttle the tree at all (instead of closing) is essentially
informative; for all intents and purposes, the tree is closed to commits
that aren't approved. If the meaning of "throttled" is unclear, we could
always just go back to having only two statuses: open or closed.

(Also, I've never seen the tree left throttled during off hours or over the
holidays. The typical situation is either an ugly open tree, or a tree
that's been closed for a long time. So I don't see how the proposed change
helps.)

- Trung

On Thu, Nov 15, 2012 at 6:03 PM, Dirk Pranke <dpra...@google.com> wrote:

> On Thu, Nov 15, 2012 at 5:56 PM, Marc-Antoine Ruel <mar...@google.com>
> wrote:
> >
> >
> > 2012/11/15 Dana Jansens <dan...@chromium.org>
> >>
> >> On Thu, Nov 15, 2012 at 8:52 PM, Dirk Pranke <dpra...@chromium.org>
> wrote:
> >> > It's been a while since I was a sheriff, but it seemed to me that the
> >> > whole point of throttling was to (a) give the bots a chance to catch
> >> > up, so that fewer changes were in flight at once and (b) help limit or
> >> > control which parts of the tree might be in flux, in order to better
> >> > triage certain kinds of failures.
> >
> >
> > (a) is keeping the tree closed.
> > (b) is also keeping the tree closed and having people commit manually.
> >
>
> I disagree. If the tree is closed, I normally interpret that as
> nothing should be landed unless it is fixing the tree; that is a
> higher bar than just throttling.
>
> >> > It may be that the CQ already checks the patches, but does it really
> >> > give us more safety than the fact that all changes are supposed to
> >> > have been run through the try servers first? (This is a real question,
> >> > maybe it does).
> >
> >
> > It does.
> >
>
> Perhaps you could say what they are?
>
> >> Especially since CQ does not run all the bots on the waterfall, such
> >> as win_aura, linux_aura.
> >
> >
> > They'll be added real soon.
> >
> >>
> >> > It seems like this defeats the point of throttling. Are there
> >> > advantages to making this change I'm not seeing?
> >
> >
> > Yes, when the tree is not closely monitored. Think of the holidays for
> > example.
> >
>
> It seems to me you're describing an open tree, not a throttled tree.
>
> -- Dirk
>
> --
> Chromium Developers mailing list: chromium-...@chromium.org
> View archives, change email options, or unsubscribe:
>     http://groups.google.com/a/chromium.org/group/chromium-dev
>

--90e6ba6138d0f658f604ce98167a
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div style=3D"font-family: arial, helvetica, sans-serif; font-size: 10pt"><=
div>+1 to Dirk&#39;s concerns.</div><div><br></div><div>As sheriff, I&#39;d=
 throttle to let some bots catch up. I might entertain commit requests to a=
 throttled tree, but I&#39;d vet them (e.g., if I&#39;m waiting for Win bot=
s to catch up, I might let a Mac-only change in).</div>
<div><br></div><div>I don&#39;t see why the CQ should get more privileges w=
ith respect to a throttled tree. A manual commit may/should also come with =
a complete set of try jobs, and moreover the committer is more likely to be=
 online and responsive!</div>
<div><br></div><div>That we throttle the tree at all (instead of closing) i=
s essentially informative; for all intents and purposes, the tree is closed=
 to commits that aren&#39;t approved. If the meaning of &quot;throttled&quo=
t; is unclear, we could always just go back to having only two statuses: op=
en or closed.</div>
<div><br></div><div>(Also, I&#39;ve never seen the tree left throttled duri=
ng off hours or over the holidays. The typical situation is either an ugly =
open tree, or a tree that&#39;s been closed for a long time. So I don&#39;t=
 see how the proposed change helps.)</div>
<div><br></div><div>- Trung</div><div><br><div class=3D"gmail_quote">On Thu=
, Nov 15, 2012 at 6:03 PM, Dirk Pranke <span dir=3D"ltr">&lt;<a href=3D"mai=
lto:dpra...@google.com" target=3D"_blank">dpra...@google.com</a>&gt;</span>=
 wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D"im">On Thu, Nov 15, 2012 at 5:=
56 PM, Marc-Antoine Ruel &lt;<a href=3D"mailto:mar...@google.com">maruel@go=
ogle.com</a>&gt; wrote:<br>

&gt;<br>
&gt;<br>
&gt; 2012/11/15 Dana Jansens &lt;<a href=3D"mailto:dan...@chromium.org">dan=
a...@chromium.org</a>&gt;<br>
&gt;&gt;<br>
&gt;&gt; On Thu, Nov 15, 2012 at 8:52 PM, Dirk Pranke &lt;<a href=3D"mailto=
:dpra...@chromium.org">dpra...@chromium.org</a>&gt; wrote:<br>
&gt;&gt; &gt; It&#39;s been a while since I was a sheriff, but it seemed to=
 me that the<br>
&gt;&gt; &gt; whole point of throttling was to (a) give the bots a chance t=
o catch<br>
&gt;&gt; &gt; up, so that fewer changes were in flight at once and (b) help=
 limit or<br>
&gt;&gt; &gt; control which parts of the tree might be in flux, in order to=
 better<br>
&gt;&gt; &gt; triage certain kinds of failures.<br>
&gt;<br>
&gt;<br>
&gt; (a) is keeping the tree closed.<br>
&gt; (b) is also keeping the tree closed and having people commit manually.=
<br>
&gt;<br>
<br>
</div>I disagree. If the tree is closed, I normally interpret that as<br>
nothing should be landed unless it is fixing the tree; that is a<br>
higher bar than just throttling.<br>
<div class=3D"im"><br>
&gt;&gt; &gt; It may be that the CQ already checks the patches, but does it=
 really<br>
&gt;&gt; &gt; give us more safety than the fact that all changes are suppos=
ed to<br>
&gt;&gt; &gt; have been run through the try servers first? (This is a real =
question,<br>
&gt;&gt; &gt; maybe it does).<br>
&gt;<br>
&gt;<br>
&gt; It does.<br>
&gt;<br>
<br>
</div>Perhaps you could say what they are?<br>
<div class=3D"im"><br>
&gt;&gt; Especially since CQ does not run all the bots on the waterfall, su=
ch<br>
&gt;&gt; as win_aura, linux_aura.<br>
&gt;<br>
&gt;<br>
&gt; They&#39;ll be added real soon.<br>
&gt;<br>
&gt;&gt;<br>
&gt;&gt; &gt; It seems like this defeats the point of throttling. Are there=
<br>
&gt;&gt; &gt; advantages to making this change I&#39;m not seeing?<br>
&gt;<br>
&gt;<br>
&gt; Yes, when the tree is not closely monitored. Think of the holidays for=
<br>
&gt; example.<br>
&gt;<br>
<br>
</div>It seems to me you&#39;re describing an open tree, not a throttled tr=
ee.<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
-- Dirk<br>
</font></span><div class=3D"HOEnZb"><div class=3D"h5"><br>
--<br>
Chromium Developers mailing list: <a href=3D"mailto:chromium-...@chromium.o=
rg">chromium-...@chromium.org</a><br>
View archives, change email options, or unsubscribe:<br>
=C2=A0 =C2=A0 <a href=3D"http://groups.google.com/a/chromium.org/group/chro=
mium-dev" target=3D"_blank">http://groups.google.com/a/chromium.org/group/c=
hromium-dev</a><br>
</div></div></blockquote></div><br></div></div>

--90e6ba6138d0f658f604ce98167a--
Create a group - Google Groups - Google Home - Terms of Service - Privacy Policy
Account Options
