Unless one could instantly spin up another ZK instance on the non-failed
frame...

+1.314.452.2896 (Tel/SMS)

On Tue, May 1, 2018 at 12:00 PM, Steph van Schalkwyk <
[email protected]> wrote:

> Adam,
> More information here:
> https://stackoverflow.com/questions/24694296/using-
> zookeeper-with-solr-but-only-have-2-servers
> Unless ZK can be instantly "reconfigured" to consider the remaining 3 as a
> full ensemble, I don't see an option.
> Steph
>
>
> +1.314.452.2896 (Tel/SMS)
>
> On Tue, May 1, 2018 at 11:53 AM, Adam Blank <[email protected]> wrote:
>
>> Thanks for your replies, Steph.  Adding back the rest of the mailing list.
>> If anyone can shed some light on my predicament that would be much
>> appreciated.
>>
>> Adam
>>
>> ---------- Forwarded message ----------
>> From: Steph van Schalkwyk <[email protected]>
>> Date: Tue, May 1, 2018 at 12:44 PM
>> Subject: Re: Configuring SolrCloud with Redundancy on Two Physical Frames
>> To: Adam Blank <[email protected]>
>>
>>
>> Maybe one of the ZK gurus could chime in? I could test it but I don't have
>> the time right now.
>>
>>
>> +1.314.452.2896 (Tel/SMS)
>>
>> On Tue, May 1, 2018 at 11:42 AM, Adam Blank <[email protected]> wrote:
>>
>> > I think I would still run into the same issue since if one frame goes
>> down
>> > I will only have 50% of ZK instances still up, and my understanding is
>> that
>> > a ZK cluster requires a majority to be up in order to operate. So if I
>> have
>> > 6 total I'd need 4 to be up.
>> >
>> >
>> > On Tue, May 1, 2018, 12:34 PM Steph van Schalkwyk <
>> [email protected]>
>> > wrote:
>> >
>> >> Only thing I can think of is to run three ZK instances on each hardware
>> >> instance. That way if one fails you still have three running on the
>> other
>> >> hardware instance.
>> >> Also, when you set up the SOLR instances, make sure you're sharding
>> >> across hardware instances, for example two shards per collection on
>> >> instance 0 and two on the other.
>> >> S
>> >>
>> >> +1.314.452.2896 (Tel/SMS)
>> >>
>> >> On Tue, May 1, 2018 at 10:17 AM, Adam Blank <[email protected]>
>> wrote:
>> >>
>> >>> Hi Steph,
>> >>>
>> >>> I should have provided some more info.  I am running on AIX.  If I'm
>> >>> understanding your comment correctly, the issue I'm having isn't with
>> being
>> >>> able to run multiple ZK instances on a single server.  The issue is
>> with
>> >>> setting up ZK and SOLR in a way that it can survive either frame
>> failure.
>> >>> I wonder if setting up a virtual IP for the ZK instances and having
>> SOLR
>> >>> connect to the VIP would work if all ZK instances share the same data
>> >>> directory on a shared drive?  I was hoping someone has encountered
>> this
>> >>> situation before, but if not, I can see if that idea would work.
>> >>>
>> >>> Thanks,
>> >>> Adam
>> >>>
>> >>> On Tue, May 1, 2018 at 10:59 AM, Steph van Schalkwyk <
>> >>> [email protected]> wrote:
>> >>>
>> >>>> Adam, is it possible to virtualize in any way?
>> >>>> As for single physical instances, I have been running three instances
>> >>>> of ZK
>> >>>> on one VM quite comfortably. This is only for dev/testing, though.
>> >>>> Regards
>> >>>> Steph
>> >>>>
>> >>>>
>> >>>> +1.314.452.2896 (Tel/SMS)
>> >>>>
>> >>>> On Tue, May 1, 2018 at 9:55 AM, Adam Blank <[email protected]>
>> >>>> wrote:
>> >>>>
>> >>>> > Hello,
>> >>>> >
>> >>>> > I would like to have a high-availability/redundant installation of
>> >>>> > Zookeeper running in my production environment. The problem is
>> that I
>> >>>> only
>> >>>> > have 2 physical frames available, so that rules out configuring a
>> >>>> Zookeeper
>> >>>> > cluster/ensemble since I'd only have redundancy if the frame with
>> the
>> >>>> > minority of servers goes down. What is the best practice in this
>> >>>> situation?
>> >>>> > Is it possible to have a separate standalone install running on
>> each
>> >>>> frame
>> >>>> > connected to the same set of SOLR nodes or to use one server as
>> >>>> primary and
>> >>>> > one as backup?
>> >>>> >
>> >>>> > Thank you,
>> >>>> > Adam
>> >>>> >
>> >>>>
>> >>>
>> >>>
>> >>
>>
>
>

Reply via email to