Mailing-List: contact reviews-help@aurora.apache.org; run by ezmlm
Precedence: bulk
Reply-To: reviews@aurora.apache.org
Content-Type: multipart/alternative;
 boundary="===============2761079484195280395=="
MIME-Version: 1.0
Subject: Re: Review Request 54957: Add option to not retry api calls to the
 scheduler.
From: Santhosh Kumar Shanmugham <santhoshkumar.s@gmail.com>
To: Santhosh Kumar Shanmugham <santhoshkumar.s@gmail.com>,
 David McLaughlin <david@dmclaughlin.com>, Joshua Cohen <jcohen@apache.org>,
 Zameer Manji <zmanji@apache.org>
Cc: Aurora ReviewBot <wfarner@apache.org>, Stephan Erb <serb@apache.org>,
 Karthik Anantha Padmanabhan <treadstone90@gmail.com>,
 Aurora <reviews@aurora.apache.org>
Date: Thu, 05 Jan 2017 22:39:59 -0000
Message-ID: <20170105223959.13478.21009@reviews.apache.org>
Auto-Submitted: auto-generated
Sender: Santhosh Kumar Shanmugham <noreply@reviews.apache.org>
References: <20170104235409.19097.20269@reviews.apache.org>
In-Reply-To: <20170104235409.19097.20269@reviews.apache.org>
Reply-To: Santhosh Kumar Shanmugham <santhoshkumar.s@gmail.com>
archived-at: Thu, 05 Jan 2017 22:40:02 -0000

--===============2761079484195280395==
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit


> On Jan. 4, 2017, 3:54 p.m., Karthik Anantha Padmanabhan wrote:
> > I was going to wrap this up - but how do people feel about making the all endpoints "idempotent" by the following method ? Inlcude an "idempotency-token" along as part of the HTTP header. This token is locally cached for, say, an hour. Every request with the same token will return simply short circuit and not be processed. The retry logic to transport layer so that we can transparently add the idempotency tokens.

Can you explain this in detail? Particularly flesh out the `HTTP headers` (aren't we using a Thrift interface while speaking to the Scheduler?) and the `locally cached` (have the Scheduler cache the `idempotency-token` or on the client?) parts.

I like this approach better, which can fix the issue across all APIs.


- Santhosh Kumar


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/54957/#review160546
-----------------------------------------------------------


On Dec. 21, 2016, 4:06 p.m., Karthik Anantha Padmanabhan wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/54957/
> -----------------------------------------------------------
> 
> (Updated Dec. 21, 2016, 4:06 p.m.)
> 
> 
> Review request for Aurora, David McLaughlin, Joshua Cohen, Santhosh Kumar Shanmugham, and Zameer Manji.
> 
> 
> Repository: aurora
> 
> 
> Description
> -------
> 
> This diff adds an option to not retry api calls to the scheduler. For some of the non-idempotent operations we would like to not automatically retry. This patch makes this functionality available only to the `schedule_backup_now` command.
> 
> If there is consensus, this can be added to all commands as well.
> 
> 
> Diffs
> -----
> 
>   src/main/python/apache/aurora/admin/admin.py 070c348d2ca5db1edecf832efd9aa5481bddaa4b 
>   src/main/python/apache/aurora/admin/aurora_admin.py fbebbab8c827b5695042d18770d850e31fc38122 
>   src/main/python/apache/aurora/client/api/__init__.py e1dde638bd1d686269fbcd88cb083a52e7f5dbfc 
>   src/main/python/apache/aurora/client/api/scheduler_client.py 9bbfece012e48e0b1752bbefd25c89e04d312cf6 
>   src/test/python/apache/aurora/client/api/test_scheduler_client.py f6018caa4f431e85a9e9ff203ac3d4b6c33f40ef 
> 
> Diff: https://reviews.apache.org/r/54957/diff/
> 
> 
> Testing
> -------
> 
> * Manuall testing
> * ./build-support/jenkins/build.sh passes
> 
> 
> Thanks,
> 
> Karthik Anantha Padmanabhan
> 
>


--===============2761079484195280395==--