I’m not talking about how much memory your machine has,
the critical bit it’s how much heap space is allocated to the
JVM to run your app.
You can increase it by specifying -Xmx2G say when you
invoke Java.
The version difference is suspicious indeed. I’m a little
confused here. Exactly _what_ program is crashing? An
independent app you wrote or nutch? If the former, you could
try compiling your Java app against the Solr jars provided
with the Solr version that ships with Nutch 1.16 (Solr 7.3.1?).
Best,
Erick
> On Jun 6, 2020, at 9:30 AM, Jim Anderson <jjanderson52...@gmail.com> wrote:
>
> Erick,
>
> Thanks for the suggestion. I will keep it in the back of my mind for now.
> My PC has 8 G-bytes of memory and has roughly 4 G-bytes in use.
>
> If the forefront, I'm looking at the recommended solr/nutch combinations.
> I'm using Solr 8.5.1 with nutch 1.16. The recommendation is to use nutch
> 1.17 with Solr 8.5.1, but 1.17 has not been released for download.
> Consequently, I used nutch 1.16. I'm not sure that will make a difference,
> but I am suspicious.
>
> Jim
>
> On Sat, Jun 6, 2020 at 9:18 AM Erick Erickson <erickerick...@gmail.com>
> wrote:
>
>> I’d look for an OutOfMemory problem before going too much farther.
>> The simplest way to see if that’s in the right direction would be to
>> run your SolrJ program with a massive memory size. Perhaps monitor
>> your program with jconsole or similar to see if there’s any clues about
>> memory usage.
>>
>> OOMs lead to unpredictable behavior, so it’s at least a possibility that
>> this is the root cause. If so, there’s nothing SolrJ can do about it
>> exactly
>> because the state of a program is indeterminate afterwards, even if the
>> OOM is caught somewhere. I suppose you could also try to catch that
>> exception in the top-level of your program.
>>
>> I’m assuming a stand-alone program here, if you’re running some custom
>> code in Solr itself, make sure the oom-killer script is running.
>>
>> Best,
>> Erick
>>
>>> On Jun 6, 2020, at 8:23 AM, Jim Anderson <jjanderson52...@gmail.com>
>> wrote:
>>>
>>> Shawn,
>>>
>>> Thanks for the explanation. Very good response.
>>>
>>> The first paragraph helped clarify what a collection is. I have read
>> quite
>>> about about Solr. There is so much to absorb that it is slowly sinking
>> in.
>>> Your 2nd paragraph definitely answered my question, i.e. passing a core
>>> name should be ok when a collection name is specified as a method
>> argument.
>>> This is what I did.
>>>
>>> Regarding the 3rd paragraph, it is good to know that Solrj is fairly
>> robust
>>> and should not be crashing. Nevertheless, that is what is happening. The
>>> call to client.query() is wrapped in a try/catch sequence. Apparently no
>>> exceptions were detected, or the program crashed before the exception
>> could
>>> be raised.
>>>
>>> My next step is to check where I can report this to the Solr folks and
>> see
>>> if they can figure out what it is crashing. BTW, I had not checked my
>>> output file before this morning. The output file indicates that the
>> program
>>> ran to completion, so I am guessing that at least one other thread is
>> being
>>> created and that that thread is crashing.
>>>
>>> Regards,
>>> Jim
>>>
>>> On Fri, Jun 5, 2020 at 10:52 PM Shawn Heisey <apa...@elyograg.org>
>> wrote:
>>>
>>>> On 6/5/2020 4:24 PM, Jim Anderson wrote:
>>>>> I am running my first solrj program and it is crashing when I call the
>>>>> method
>>>>>
>>>>> client.query("coreName",queryParms)
>>>>>
>>>>> The API doc says the string should be a collection. I'm still not sure
>>>>> about the difference between a collection and a core, so what I am
>> doing
>>>> is
>>>>> likely illegal. Given that I have created a core, create a collection
>>>> from
>>>>> it so that I can truly pass a collection name to the query function?
>>>>
>>>> The concept of a collection comes from SolrCloud. A collection is made
>>>> up of one or more shards. A shard is made up of one or more replicas.
>>>> Each replica is a core. If you're not running SolrCloud, then you do
>>>> not have collections.
>>>>
>>>> Wherever SolrJ docs says "collection" as a parameter for a request, it
>>>> is likely that you can think "core" instead and have it still be
>>>> correct. If you're running SolrCloud, you'll want to be very careful to
>>>> know the difference.
>>>>
>>>> It seems very odd for a SolrJ query to cause the program to crash. It
>>>> would be pretty common for it to throw an exception, but that's not the
>>>> same as a crash, unless exception handling is incorrect or missing.
>>>>
>>>> Thanks,
>>>> Shawn
>>>>
>>
>>
