You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by Anderson vasconcelos <an...@gmail.com> on 2012/01/24 20:36:01 UTC
Indexing failover and replication
Hi
I'm doing now a test with replication using solr 1.4.1. I configured
two servers (server1 and server 2) as master/slave to sincronized
both. I put apache on the front side, and we index sometime in server1
and sometime in server2.
I realized that the both index servers are now confused. In solr data
folder, was created many index folders with the timestamp of
syncronization (Exemple: index.20120124041340) with some segments
inside.
I thought that was possible to index in two master server and than
synchronized both using replication. It's really possible do this with
replication mechanism? If is possible, what I have done wrong?
I need to have more than one node for indexing to guarantee failover
feature for indexing. MultiMaster is the best way to guarantee
failover feature for indexing?
Thanks
Re: Indexing failover and replication
Posted by Lance Norskog <go...@gmail.com>.
You could just have each Solr index and query from its own index, and
not copy indexes.
On Wed, Jan 25, 2012 at 11:24 AM, Anderson vasconcelos
<an...@gmail.com> wrote:
> Thanks for the Reply Erick
> I will make the replication to both master manually.
>
> Thanks
>
> 2012/1/25, Erick Erickson <er...@gmail.com>:
>> No, there no good ways to have a single slave know about
>> two masters and just use the right one. It sounds like you've
>> got each machine being both a master and a slave? This is
>> not supported. What you probably want to do is either set
>> up a repeater or just index to the two masters and manually
>> change the back to the primary if the primary goes down, having
>> all replication happen from the master.
>>
>> Best
>> Erick
>>
>> On Tue, Jan 24, 2012 at 11:36 AM, Anderson vasconcelos
>> <an...@gmail.com> wrote:
>>> Hi
>>> I'm doing now a test with replication using solr 1.4.1. I configured
>>> two servers (server1 and server 2) as master/slave to sincronized
>>> both. I put apache on the front side, and we index sometime in server1
>>> and sometime in server2.
>>>
>>> I realized that the both index servers are now confused. In solr data
>>> folder, was created many index folders with the timestamp of
>>> syncronization (Exemple: index.20120124041340) with some segments
>>> inside.
>>>
>>> I thought that was possible to index in two master server and than
>>> synchronized both using replication. It's really possible do this with
>>> replication mechanism? If is possible, what I have done wrong?
>>>
>>> I need to have more than one node for indexing to guarantee failover
>>> feature for indexing. MultiMaster is the best way to guarantee
>>> failover feature for indexing?
>>>
>>> Thanks
>>
--
Lance Norskog
goksron@gmail.com
Re: Indexing failover and replication
Posted by Anderson vasconcelos <an...@gmail.com>.
Thanks for the Reply Erick
I will make the replication to both master manually.
Thanks
2012/1/25, Erick Erickson <er...@gmail.com>:
> No, there no good ways to have a single slave know about
> two masters and just use the right one. It sounds like you've
> got each machine being both a master and a slave? This is
> not supported. What you probably want to do is either set
> up a repeater or just index to the two masters and manually
> change the back to the primary if the primary goes down, having
> all replication happen from the master.
>
> Best
> Erick
>
> On Tue, Jan 24, 2012 at 11:36 AM, Anderson vasconcelos
> <an...@gmail.com> wrote:
>> Hi
>> I'm doing now a test with replication using solr 1.4.1. I configured
>> two servers (server1 and server 2) as master/slave to sincronized
>> both. I put apache on the front side, and we index sometime in server1
>> and sometime in server2.
>>
>> I realized that the both index servers are now confused. In solr data
>> folder, was created many index folders with the timestamp of
>> syncronization (Exemple: index.20120124041340) with some segments
>> inside.
>>
>> I thought that was possible to index in two master server and than
>> synchronized both using replication. It's really possible do this with
>> replication mechanism? If is possible, what I have done wrong?
>>
>> I need to have more than one node for indexing to guarantee failover
>> feature for indexing. MultiMaster is the best way to guarantee
>> failover feature for indexing?
>>
>> Thanks
>
Re: Indexing failover and replication
Posted by Erick Erickson <er...@gmail.com>.
No, there no good ways to have a single slave know about
two masters and just use the right one. It sounds like you've
got each machine being both a master and a slave? This is
not supported. What you probably want to do is either set
up a repeater or just index to the two masters and manually
change the back to the primary if the primary goes down, having
all replication happen from the master.
Best
Erick
On Tue, Jan 24, 2012 at 11:36 AM, Anderson vasconcelos
<an...@gmail.com> wrote:
> Hi
> I'm doing now a test with replication using solr 1.4.1. I configured
> two servers (server1 and server 2) as master/slave to sincronized
> both. I put apache on the front side, and we index sometime in server1
> and sometime in server2.
>
> I realized that the both index servers are now confused. In solr data
> folder, was created many index folders with the timestamp of
> syncronization (Exemple: index.20120124041340) with some segments
> inside.
>
> I thought that was possible to index in two master server and than
> synchronized both using replication. It's really possible do this with
> replication mechanism? If is possible, what I have done wrong?
>
> I need to have more than one node for indexing to guarantee failover
> feature for indexing. MultiMaster is the best way to guarantee
> failover feature for indexing?
>
> Thanks