On Wed, Jul 8, 2009 at 10:14 PM, solr jay<solr...@gmail.com> wrote:
> Thanks. The patch looks good, and I now see the new index directory and it
> is in sync with the one on master. I'll do more testing.
>
> It is probably not important, but I am just curious why we switch index
> directory. I thought it would be easier to just rename index to index.*, and
> rename the new index directory to index.
It is for consistency across OS's . Windows would not let me do a rename.
>
> 2009/7/7 Noble Paul നോബിള് नोब्ळ् <noble.p...@corp.aol.com>
>>
>> jay,
>> Thanks. The testcase was not enough. I have given a new patch . I
>> guess that should solve this
>>
>> On Wed, Jul 8, 2009 at 3:48 AM, solr jay<solr...@gmail.com> wrote:
>> > I guess in this case it doesn't matter whether the two directories
>> > tmpIndexDir and indexDir are the same or not. It looks that the index
>> > directory is switched to tmpIndexDir and then it is deleted inside
>> > "finally".
>> >
>> > On Tue, Jul 7, 2009 at 12:31 PM, solr jay <solr...@gmail.com> wrote:
>> >
>> >> In fact, I saw the directory was created and then deleted.
>> >>
>> >>
>> >> On Tue, Jul 7, 2009 at 12:29 PM, solr jay <solr...@gmail.com> wrote:
>> >>
>> >>> Ok, Here is the problem. In the function, the two directories
>> >>> tmpIndexDir
>> >>> and indexDir are the same (in this case only?), and then at the end of
>> >>> the
>> >>> function, the directory tmpIndexDir is deleted, which deletes the new
>> >>> index
>> >>> directory.
>> >>>
>> >>>
>> >>> } finally {
>> >>> delTree(tmpIndexDir);
>> >>>
>> >>> }
>> >>>
>> >>>
>> >>> On Tue, Jul 7, 2009 at 12:17 PM, solr jay <solr...@gmail.com> wrote:
>> >>>
>> >>>> I see. So I tried it again. Now index.properties has
>> >>>>
>> >>>> #index properties
>> >>>> #Tue Jul 07 12:13:49 PDT 2009
>> >>>> index=index.20090707121349
>> >>>>
>> >>>> but there is no such directory index.20090707121349 under the data
>> >>>> directory.
>> >>>>
>> >>>> Thanks,
>> >>>>
>> >>>> J
>> >>>>
>> >>>>
>> >>>> On Tue, Jul 7, 2009 at 11:50 AM, Shalin Shekhar Mangar <
>> >>>> shalinman...@gmail.com> wrote:
>> >>>>
>> >>>>> On Tue, Jul 7, 2009 at 11:50 PM, solr jay <solr...@gmail.com> wrote:
>> >>>>>
>> >>>>> > It seemed that the patch fixed the symptom, but not the problem
>> >>>>> itself.
>> >>>>> >
>> >>>>> > Now the log messages looks good. After one download and installed
>> >>>>> > the
>> >>>>> > index,
>> >>>>> > it printed out
>> >>>>> >
>> >>>>> > *Jul 7, 2009 10:35:10 AM org.apache.solr.handler.SnapPuller
>> >>>>> > fetchLatestIndex
>> >>>>> > INFO: Slave in sync with master.*
>> >>>>> >
>> >>>>> > but the files inside index directory did not change. Both
>> >>>>> index.properties
>> >>>>> > and replication.properties were updated though.
>> >>>>> >
>> >>>>>
>> >>>>> Note that in this case, Solr would have created a new index
>> >>>>> directory.
>> >>>>> Are
>> >>>>> you comparing the files on the slave in the new index directory? You
>> >>>>> can
>> >>>>> get
>> >>>>> the new index directory's name from index.properties.
>> >>>>>
>> >>>>> --
>> >>>>> Regards,
>> >>>>> Shalin Shekhar Mangar.
>> >>>>>
>> >>>>
>> >>>>
>> >>>
>> >>
>> >
>> >
>> > --
>> > J
>> >
>>
>>
>>
>> --
>> -----------------------------------------------------
>> Noble Paul | Principal Engineer| AOL | http://aol.com
>
>
>
> --
> J
>
--
-----------------------------------------------------
Noble Paul | Principal Engineer| AOL | http://aol.com
