Disable QUIC port selection when the server is going away.

Disable QUIC port selection when the server is going away.
There is no point in trying to return to the same server, if that server
is no longer handling requests.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=255293

[jar (doing other things), 2014-03-04 01:26:11]

This seems like a strange hack.  I can justify it in some cases, but not in
others.

If you want the go-away to include "avoid using the same port" I think we should
say so in the go-away (as some sort of option).  Similarly, if the go-away would
like us to use a different Connection ID (because it "knows" that said ID will
hit a different machine), then that should be stated in the go-away.  Finally,
if the go-away would like some period of non-QUIC usage (i.e., temporary disable
of alt-protocol QUIC), it would be nice to say that.

It would probably be nice to indicate if the crypto config should be
"speculatively reused" for a reconnection (and hence support a 0-RTT), or if the
reconnection required an inchoate CHLO.

[alyssar, 2014-03-05 14:07:13]

Could you explain why you find it strange?   If you get a GOAWAY it means
the server doesn't want you to reconnect to it (going away, overloaded
etc.).  If so, it makes a whole lot of sense to me to avoid logic we wrote
specifically to try to reconnect to the same machine.


On Mon, Mar 3, 2014 at 8:26 PM, <jar@chromium.org> wrote:

> This seems like a strange hack.  I can justify it in some cases, but not in
> others.
>
> If you want the go-away to include "avoid using the same port" I think we
> should
> say so in the go-away (as some sort of option).  Similarly, if the go-away
> would
> like us to use a different Connection ID (because it "knows" that said ID
> will
> hit a different machine), then that should be stated in the go-away.
>  Finally,
> if the go-away would like some period of non-QUIC usage (i.e., temporary
> disable
> of alt-protocol QUIC), it would be nice to say that.
>
> It would probably be nice to indicate if the crypto config should be
> "speculatively reused" for a reconnection (and hence support a 0-RTT), or
> if the
> reconnection required an inchoate CHLO.
>
> https://codereview.chromium.org/186293002/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[jar (doing other things), 2014-03-05 19:52:46]

Per offline discussion, this is meant to be a short term hack to better deal
with the go-away issues as servers are rebooted.

I'm hopeful that we'll revisit this in the future, with increased info in the
go-away message (rather than assuming that the "right thing" to do is to
reconnect using a different (enough?) port" (which is what this patch gets us).

LGTM

[Ryan Hamilton, 2014-03-05 19:57:04]

On 2014/03/05 19:52:46, jar wrote:
> Per offline discussion, this is meant to be a short term hack to better deal
> with the go-away issues as servers are rebooted.
> 
> I'm hopeful that we'll revisit this in the future, with increased info in the
> go-away message (rather than assuming that the "right thing" to do is to
> reconnect using a different (enough?) port" (which is what this patch gets
us).
> 
> LGTM

Thanks, jar!  Landing...

[Ryan Hamilton, 2014-03-05 19:57:08]

The CQ bit was checked by rch@chromium.org

[commit-bot: I haz the power, 2014-03-05 19:58:17]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/rch@chromium.org/186293002/40001

[commit-bot: I haz the power, 2014-03-05 22:02:02]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-03-05 22:02:03]

Retried try job too often on win_rel for step(s) unit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[Ryan Hamilton, 2014-03-05 22:15:04]

The CQ bit was checked by rch@chromium.org

[commit-bot: I haz the power, 2014-03-05 22:18:06]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/rch@chromium.org/186293002/40001

[commit-bot: I haz the power, 2014-03-06 00:06:10]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/rch@chromium.org/186293002/40001

[commit-bot: I haz the power, 2014-03-06 02:06:01]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-03-06 02:06:02]

Retried try job too often on win_rel for step(s) unit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[Ryan Hamilton, 2014-03-06 05:18:39]

The CQ bit was checked by rch@chromium.org

[commit-bot: I haz the power, 2014-03-06 05:19:07]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/rch@chromium.org/186293002/40001

[commit-bot: I haz the power, 2014-03-06 10:30:11]

Change committed as 255293