Retry fetching patch from rietveld on a 404

Retry fetching patch from rietveld on a 404

Also added a tiny bit of exponential backoff.

BUG=375479
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=291682

[M-A Ruel, 2014-06-21 01:40:33]

I think it's too deep to do this kind of retries. We don't want it to retry
arbitrary endpoints on 404. Please do the retry at an higher level, e.g. in
rietveld.py.

[Ryan Tseng, 2014-06-30 20:37:25]

Done, tested by apply_issue --issue 350433003 in depot_tools.

[hinoka, 2014-07-28 21:08:01]

gengle ping

[hinoka, 2014-07-28 21:08:07]

gentle*

[M-A Ruel, 2014-07-28 21:11:39]

https://codereview.chromium.org/350433003/diff/40001/rietveld.py
File rietveld.py (right):

https://codereview.chromium.org/350433003/diff/40001/rietveld.py#newcode103
rietveld.py:103: data = json.loads(self.get(url))
It'd be more sane to add a retry_404=True flag to _send() function instead of
two loops of retries on over the other.

[hinoka, 2014-08-27 13:42:47]

Actually cleaning out lower-priority git bugs now.  1-month delay ping.

https://codereview.chromium.org/350433003/diff/40001/rietveld.py
File rietveld.py (right):

https://codereview.chromium.org/350433003/diff/40001/rietveld.py#newcode103
rietveld.py:103: data = json.loads(self.get(url))
On 2014/07/28 21:11:38, M-A Ruel wrote:
> It'd be more sane to add a retry_404=True flag to _send() function instead of
> two loops of retries on over the other.

Done.

[M-A Ruel, 2014-08-27 13:54:25]

lgtm

[Ryan Tseng, 2014-08-27 13:58:13]

The CQ bit was checked by hinoka@google.com

[commit-bot: I haz the power, 2014-08-27 13:58:47]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/hinoka@chromium.org/350433003/60001

[commit-bot: I haz the power, 2014-08-27 14:00:44]

Committed patchset #4 (60001) as 291682

[robwu, 2014-10-26 09:43:47]

What is the reason for retrying up to 40 times when Rietveld replies with 404 (I
can't see the linked bug)?

I ran "git cl status" and thought that something went wrong due to the lack of
feedback... It would have been really nice if the tool warned that an issue
could not be found, and suggest to run "git cl status 0" to reset the issue for
the branch. Now I get zero feedback and discovered the cause only after
following the execution path in git_cl.py and rietveld.py.

[Dirk Pranke, 2014-10-26 15:46:28]

On 2014/10/26 09:43:47, robwu wrote:
> What is the reason for retrying up to 40 times when Rietveld replies with 404
(I
> can't see the linked bug)?

I cleared the restrict-view label from the bug; there was no reason for it to be
restricted.

My guess is that the theory was that the 404 was a result of Rietveld being
flaky in a "eventually-but-not-yet-consistent" way.