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 Bernd Fehling <be...@uni-bielefeld.de> on 2013/02/01 10:13:32 UTC
expert question about SolrReplication
A question to the experts,
why is the replicated index copied from its temporary location (index.xxxxxxxxx)
to the real index directory and NOT moved?
Copying over 100s of gigs takes some time, moving is just changing the file system link.
Also, instead of first deleting the old index, why not
- moving the file links of old index to "index.xxxxx.old"
- moving the file links of new index to "index"
- and finally after new searcher is up, deleting "index.xxxxx.old"
Any answers?
Regards
Bernd
Re: expert question about SolrReplication
Posted by Bernd Fehling <be...@uni-bielefeld.de>.
Am 02.02.2013 03:48, schrieb Yonik Seeley:
> On Fri, Feb 1, 2013 at 4:13 AM, Bernd Fehling
> <be...@uni-bielefeld.de> wrote:
>> A question to the experts,
>>
>> why is the replicated index copied from its temporary location (index.xxxxxxxxx)
>> to the real index directory and NOT moved?
>
> The intent is certainly to move and not copy (provided the Directory
> supports it).
> See StandardDirectoryFactory.move()
Because running Solr/Lucene with Linux I suppose it should really move,
but I will step through it with debugger and see what happens.
>
>> Copying over 100s of gigs takes some time, moving is just changing the file system link.
>>
>> Also, instead of first deleting the old index, why not
>> - moving the file links of old index to "index.xxxxx.old"
>
> You can't do this in Windows?
Solr/Lucene is optimized for Windows???
Who is on the MS payroll?
>
>> - moving the file links of new index to "index"
>> - and finally after new searcher is up, deleting "index.xxxxx.old"
>>
>
> -Yonik
> http://lucidworks.com
>
Re: expert question about SolrReplication
Posted by Yonik Seeley <yo...@lucidworks.com>.
On Fri, Feb 1, 2013 at 4:13 AM, Bernd Fehling
<be...@uni-bielefeld.de> wrote:
> A question to the experts,
>
> why is the replicated index copied from its temporary location (index.xxxxxxxxx)
> to the real index directory and NOT moved?
The intent is certainly to move and not copy (provided the Directory
supports it).
See StandardDirectoryFactory.move()
> Copying over 100s of gigs takes some time, moving is just changing the file system link.
>
> Also, instead of first deleting the old index, why not
> - moving the file links of old index to "index.xxxxx.old"
You can't do this in Windows?
> - moving the file links of new index to "index"
> - and finally after new searcher is up, deleting "index.xxxxx.old"
>
-Yonik
http://lucidworks.com