You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@airavata.apache.org by Lahiru Ginnaliya Gamathige <gl...@gmail.com> on 2015/04/21 18:39:59 UTC
Replacing ZK client with Curator
Hi Devs,
I think its a good idea to replace ZK client code with Apache Curator
because we have to handle lots of timeout and other connectionLost issues
manually.
What do you think, is it too late to do this and do the release ? If so we
can do it in next release otherwise we can implement this and extend the
release bit further.
Are there any blocker like issues cause due to ZK timeout and
connectionLost errors ?
Regards
Lahiru
--
Research Assistant
Science Gateways Group
Indiana University
Re: Replacing ZK client with Curator
Posted by Suresh Marru <sm...@apache.org>.
Hi Lahiru,
I only played with it very little, I was prototyping multiple API services and used curator service discovery recipe. I liked it from a development standpoint, do not have a opinion on how well it will help mitigate the issues you mention below. Hypothetically yes, but not sure in practice.
Should we just try it in a branch and validate it with the test framework?
If it helps, I can bring this thread to attention of the curator pmc and seek expert advice, do we need it?
Suresh
> On Apr 21, 2015, at 1:34 PM, Lahiru Ginnaliya Gamathige <gl...@gmail.com> wrote:
>
> Its not replace one set of problems with another, its about Curator is handling timeouts and other ZK client code we have to do, so we can directly use the Curator code. Its like a helper library written using vanilla ZK client code.
>
> On Tue, Apr 21, 2015 at 1:30 PM, Marlon Pierce <marpierc@iu.edu <ma...@iu.edu>> wrote:
> Have you done a preliminary evaluation? We don't want to replace one set of problems with another.
>
> Marlon
>
>
> On 4/21/15 12:39 PM, Lahiru Ginnaliya Gamathige wrote:
> Hi Devs,
>
> I think its a good idea to replace ZK client code with Apache Curator
> because we have to handle lots of timeout and other connectionLost issues
> manually.
>
> What do you think, is it too late to do this and do the release ? If so we
> can do it in next release otherwise we can implement this and extend the
> release bit further.
>
> Are there any blocker like issues cause due to ZK timeout and
> connectionLost errors ?
>
> Regards
> Lahiru
>
>
>
>
>
> --
> Research Assistant
> Science Gateways Group
> Indiana University
Re: Replacing ZK client with Curator
Posted by Shameera Rathnayaka <sh...@gmail.com>.
Hi Devs,
I replaced zookeeper client with curator and ran few tests on real compute
resources, all went well. And tested recovery scenario too. I don't see
connection issues at all after integrate apache curator which we were
seeing time to time in logs. I will merge curator integration branch to
master.
Thanks,
Shameera.
On Tue, Apr 21, 2015 at 1:45 PM, Chathuri Wimalasena <ka...@gmail.com>
wrote:
> +1 for replacing with apache curator. I updated dev server after
> yesterday's updates. Eroma ran her test scripts and yesterday and I can see
> some jobs are running today as well. I don't see any blockers related to ZK
> timeout and connection losses.
>
> On Tue, Apr 21, 2015 at 1:34 PM, Lahiru Ginnaliya Gamathige <
> glahiru@gmail.com> wrote:
>
>> Its not replace one set of problems with another, its about Curator is
>> handling timeouts and other ZK client code we have to do, so we can
>> directly use the Curator code. Its like a helper library written using
>> vanilla ZK client code.
>>
>> On Tue, Apr 21, 2015 at 1:30 PM, Marlon Pierce <ma...@iu.edu> wrote:
>>
>>> Have you done a preliminary evaluation? We don't want to replace one
>>> set of problems with another.
>>>
>>> Marlon
>>>
>>>
>>> On 4/21/15 12:39 PM, Lahiru Ginnaliya Gamathige wrote:
>>>
>>>> Hi Devs,
>>>>
>>>> I think its a good idea to replace ZK client code with Apache Curator
>>>> because we have to handle lots of timeout and other connectionLost
>>>> issues
>>>> manually.
>>>>
>>>> What do you think, is it too late to do this and do the release ? If so
>>>> we
>>>> can do it in next release otherwise we can implement this and extend the
>>>> release bit further.
>>>>
>>>> Are there any blocker like issues cause due to ZK timeout and
>>>> connectionLost errors ?
>>>>
>>>> Regards
>>>> Lahiru
>>>>
>>>>
>>>
>>
>>
>> --
>> Research Assistant
>> Science Gateways Group
>> Indiana University
>>
>
>
--
Best Regards,
Shameera Rathnayaka.
email: shameera AT apache.org , shameerainfo AT gmail.com
Blog : http://shameerarathnayaka.blogspot.com/
Re: Replacing ZK client with Curator
Posted by Chathuri Wimalasena <ka...@gmail.com>.
+1 for replacing with apache curator. I updated dev server after
yesterday's updates. Eroma ran her test scripts and yesterday and I can see
some jobs are running today as well. I don't see any blockers related to ZK
timeout and connection losses.
On Tue, Apr 21, 2015 at 1:34 PM, Lahiru Ginnaliya Gamathige <
glahiru@gmail.com> wrote:
> Its not replace one set of problems with another, its about Curator is
> handling timeouts and other ZK client code we have to do, so we can
> directly use the Curator code. Its like a helper library written using
> vanilla ZK client code.
>
> On Tue, Apr 21, 2015 at 1:30 PM, Marlon Pierce <ma...@iu.edu> wrote:
>
>> Have you done a preliminary evaluation? We don't want to replace one set
>> of problems with another.
>>
>> Marlon
>>
>>
>> On 4/21/15 12:39 PM, Lahiru Ginnaliya Gamathige wrote:
>>
>>> Hi Devs,
>>>
>>> I think its a good idea to replace ZK client code with Apache Curator
>>> because we have to handle lots of timeout and other connectionLost issues
>>> manually.
>>>
>>> What do you think, is it too late to do this and do the release ? If so
>>> we
>>> can do it in next release otherwise we can implement this and extend the
>>> release bit further.
>>>
>>> Are there any blocker like issues cause due to ZK timeout and
>>> connectionLost errors ?
>>>
>>> Regards
>>> Lahiru
>>>
>>>
>>
>
>
> --
> Research Assistant
> Science Gateways Group
> Indiana University
>
Re: Replacing ZK client with Curator
Posted by Lahiru Ginnaliya Gamathige <gl...@gmail.com>.
Its not replace one set of problems with another, its about Curator is
handling timeouts and other ZK client code we have to do, so we can
directly use the Curator code. Its like a helper library written using
vanilla ZK client code.
On Tue, Apr 21, 2015 at 1:30 PM, Marlon Pierce <ma...@iu.edu> wrote:
> Have you done a preliminary evaluation? We don't want to replace one set
> of problems with another.
>
> Marlon
>
>
> On 4/21/15 12:39 PM, Lahiru Ginnaliya Gamathige wrote:
>
>> Hi Devs,
>>
>> I think its a good idea to replace ZK client code with Apache Curator
>> because we have to handle lots of timeout and other connectionLost issues
>> manually.
>>
>> What do you think, is it too late to do this and do the release ? If so we
>> can do it in next release otherwise we can implement this and extend the
>> release bit further.
>>
>> Are there any blocker like issues cause due to ZK timeout and
>> connectionLost errors ?
>>
>> Regards
>> Lahiru
>>
>>
>
--
Research Assistant
Science Gateways Group
Indiana University
Re: Replacing ZK client with Curator
Posted by Marlon Pierce <ma...@iu.edu>.
Have you done a preliminary evaluation? We don't want to replace one
set of problems with another.
Marlon
On 4/21/15 12:39 PM, Lahiru Ginnaliya Gamathige wrote:
> Hi Devs,
>
> I think its a good idea to replace ZK client code with Apache Curator
> because we have to handle lots of timeout and other connectionLost issues
> manually.
>
> What do you think, is it too late to do this and do the release ? If so we
> can do it in next release otherwise we can implement this and extend the
> release bit further.
>
> Are there any blocker like issues cause due to ZK timeout and
> connectionLost errors ?
>
> Regards
> Lahiru
>