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 Dmitry Kan <so...@gmail.com> on 2014/02/03 13:03:36 UTC
weird exception on update
Hello!
We are hitting a really strange and nasty issue when trying to delete by
query and not when just adding documents. The exception says:
http://pastebin.com/B1x5dAF7
Any ideas as to what is going on?
The delete by query is referencing the unique field. The core's index does
not contain the value that is being deleted.
Solr: 4.3.1.
--
Dmitry
Blog: http://dmitrykan.blogspot.com
Twitter: twitter.com/dmitrykan
Re: weird exception on update
Posted by Dmitry Kan <so...@gmail.com>.
Hi Hoss,
Thanks for replying. I have created a jira:
https://issues.apache.org/jira/browse/SOLR-5697
It contains the required configs (actually a shard) and a query parser
maven project. These illustrate the issue.
I had to omit the solr.war from the webapps of the shard as it exceeded the
upload limit. Hope it is not a problem for you to fetch and deploy that.
Please let me know, if the jira is any clearer or if more detail is needed.
Thanks,
Dmitry
On Wed, Feb 5, 2014 at 2:58 AM, Chris Hostetter <ho...@fucit.org>wrote:
>
> : > We found out that:
> : >
> : > 1. this happens iff on two cores inside the same container there is a
> : > query parser defined via defType.
> : > 2. After removing index files on one of the cores, the delete by query
> : > works just fine. Right after restarting the container, the same query
> fails.
> ...
> : The solution (or workaround?) is to drop the defType from one of the
> cores
> : and use {!qparser} local param on every query, including the delete by
> : query. It would be really great, if this could be handled on the solr
> : config side only without involving the client changes.
>
> Dimitry: I'm having a really hard time wrapping my head arround what
> exactly you are describing.
>
> can you please either:
> 1) open a jira issue containing a patch with a test that demonstrates the
> problem
> 2) describe in detail the minimum steps someone (like me) would need to
> take in order to reproduce your problem -- either including a full set of
> configs & data, or some instructions on how to modify the sample configs /
> data
>
> (ie: if the bug only reproduces with multiple cores, then maybe some
> instructions on what changes to make to the configs
> in solr/example/multicore, and then what data to load, and what deletes to
> run, etc..)
>
>
>
>
> -Hoss
> http://www.lucidworks.com/
>
--
Dmitry
Blog: http://dmitrykan.blogspot.com
Twitter: twitter.com/dmitrykan
Re: weird exception on update
Posted by Chris Hostetter <ho...@fucit.org>.
: > We found out that:
: >
: > 1. this happens iff on two cores inside the same container there is a
: > query parser defined via defType.
: > 2. After removing index files on one of the cores, the delete by query
: > works just fine. Right after restarting the container, the same query fails.
...
: The solution (or workaround?) is to drop the defType from one of the cores
: and use {!qparser} local param on every query, including the delete by
: query. It would be really great, if this could be handled on the solr
: config side only without involving the client changes.
Dimitry: I'm having a really hard time wrapping my head arround what
exactly you are describing.
can you please either:
1) open a jira issue containing a patch with a test that demonstrates the
problem
2) describe in detail the minimum steps someone (like me) would need to
take in order to reproduce your problem -- either including a full set of
configs & data, or some instructions on how to modify the sample configs /
data
(ie: if the bug only reproduces with multiple cores, then maybe some
instructions on what changes to make to the configs
in solr/example/multicore, and then what data to load, and what deletes to
run, etc..)
-Hoss
http://www.lucidworks.com/
Re: weird exception on update
Posted by Dmitry Kan <so...@gmail.com>.
We are still hitting an issue with two cores, each having their own custom
query parser.
The problem in passing {!qparser} is that the custom query parser can
pretty much alter an input query into something that is not desirable for
the delete by query operation.
Is there any way of specifying explicitly lucene's query parser for the
delete by query? If no query parser is specified in such a config we get an
error:
<?xml version="1.0" encoding="UTF-8"?>
<response>
<lst name="responseHeader"><int name="status">400</int><int
name="QTime">34</int></lst><lst name="error"><str name="msg">Unknown query
parser 'lucene'</str><int name="code">400</int></lst>
</response>
Any ideas are welcome.
Dmitry
On Mon, Feb 3, 2014 at 4:49 PM, Dmitry Kan <so...@gmail.com> wrote:
> The solution (or workaround?) is to drop the defType from one of the cores
> and use {!qparser} local param on every query, including the delete by
> query. It would be really great, if this could be handled on the solr
> config side only without involving the client changes.
>
>
>
>
> On Mon, Feb 3, 2014 at 4:02 PM, Dmitry Kan <so...@gmail.com> wrote:
>
>> This exception is similar to what is talked about here:
>> https://gist.github.com/mbklein/6367133
>> http://irc.projecthydra.org/2013-08-28.html
>>
>> We found out that:
>>
>> 1. this happens iff on two cores inside the same container there is a
>> query parser defined via defType.
>> 2. After removing index files on one of the cores, the delete by query
>> works just fine. Right after restarting the container, the same query fails.
>>
>>
>> Is there a jira for this? Should I create one?
>>
>> Dmitry
>>
>>
>> On Mon, Feb 3, 2014 at 2:03 PM, Dmitry Kan <so...@gmail.com> wrote:
>>
>>> Hello!
>>>
>>> We are hitting a really strange and nasty issue when trying to delete by
>>> query and not when just adding documents. The exception says:
>>> http://pastebin.com/B1x5dAF7
>>>
>>> Any ideas as to what is going on?
>>>
>>> The delete by query is referencing the unique field. The core's index
>>> does not contain the value that is being deleted.
>>> Solr: 4.3.1.
>>>
>>>
>>> --
>>> Dmitry
>>> Blog: http://dmitrykan.blogspot.com
>>> Twitter: twitter.com/dmitrykan
>>>
>>
>>
>>
>> --
>> Dmitry
>> Blog: http://dmitrykan.blogspot.com
>> Twitter: twitter.com/dmitrykan
>>
>
>
>
> --
> Dmitry
> Blog: http://dmitrykan.blogspot.com
> Twitter: twitter.com/dmitrykan
>
--
Dmitry
Blog: http://dmitrykan.blogspot.com
Twitter: twitter.com/dmitrykan
Re: weird exception on update
Posted by Dmitry Kan <so...@gmail.com>.
The solution (or workaround?) is to drop the defType from one of the cores
and use {!qparser} local param on every query, including the delete by
query. It would be really great, if this could be handled on the solr
config side only without involving the client changes.
On Mon, Feb 3, 2014 at 4:02 PM, Dmitry Kan <so...@gmail.com> wrote:
> This exception is similar to what is talked about here:
> https://gist.github.com/mbklein/6367133
> http://irc.projecthydra.org/2013-08-28.html
>
> We found out that:
>
> 1. this happens iff on two cores inside the same container there is a
> query parser defined via defType.
> 2. After removing index files on one of the cores, the delete by query
> works just fine. Right after restarting the container, the same query fails.
>
>
> Is there a jira for this? Should I create one?
>
> Dmitry
>
>
> On Mon, Feb 3, 2014 at 2:03 PM, Dmitry Kan <so...@gmail.com> wrote:
>
>> Hello!
>>
>> We are hitting a really strange and nasty issue when trying to delete by
>> query and not when just adding documents. The exception says:
>> http://pastebin.com/B1x5dAF7
>>
>> Any ideas as to what is going on?
>>
>> The delete by query is referencing the unique field. The core's index
>> does not contain the value that is being deleted.
>> Solr: 4.3.1.
>>
>>
>> --
>> Dmitry
>> Blog: http://dmitrykan.blogspot.com
>> Twitter: twitter.com/dmitrykan
>>
>
>
>
> --
> Dmitry
> Blog: http://dmitrykan.blogspot.com
> Twitter: twitter.com/dmitrykan
>
--
Dmitry
Blog: http://dmitrykan.blogspot.com
Twitter: twitter.com/dmitrykan
Re: weird exception on update
Posted by Dmitry Kan <so...@gmail.com>.
This exception is similar to what is talked about here:
https://gist.github.com/mbklein/6367133
http://irc.projecthydra.org/2013-08-28.html
We found out that:
1. this happens iff on two cores inside the same container there is a query
parser defined via defType.
2. After removing index files on one of the cores, the delete by query
works just fine. Right after restarting the container, the same query fails.
Is there a jira for this? Should I create one?
Dmitry
On Mon, Feb 3, 2014 at 2:03 PM, Dmitry Kan <so...@gmail.com> wrote:
> Hello!
>
> We are hitting a really strange and nasty issue when trying to delete by
> query and not when just adding documents. The exception says:
> http://pastebin.com/B1x5dAF7
>
> Any ideas as to what is going on?
>
> The delete by query is referencing the unique field. The core's index does
> not contain the value that is being deleted.
> Solr: 4.3.1.
>
>
> --
> Dmitry
> Blog: http://dmitrykan.blogspot.com
> Twitter: twitter.com/dmitrykan
>
--
Dmitry
Blog: http://dmitrykan.blogspot.com
Twitter: twitter.com/dmitrykan