Re: Error while upgrading from 1.4 to 3.1
well fixed for now... just ignore this thread - Zeki ama calismiyor... Calissa yapar... -- View this message in context: http://lucene.472066.n3.nabble.com/Error-while-upgrading-from-1-4-to-3-1-tp3492373p3492887.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: representing latlontype in pojo
Am 08.11.2011 23:38, schrieb Cam Bazz: How can I store a 2d point and index it to a field type that is latlontype, if I am using solrj? Simply use a String field. The format is "$latitude,$longitude". -Kuli
Re: Solr dismax scoring and weight
Thanks for the details, but what do you mean by normalization, can you describe shortly the concepts behind ? -- View this message in context: http://lucene.472066.n3.nabble.com/Solr-dismax-scoring-and-weight-tp3490096p3492986.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: Search Correlated Data between Multivalued Fields
I do not think this is possbile directly out of the box in Solr. A quick workaround would be to fully denormalize the data, ie instead of multivalued notes for a customer, have a completely flat index of customer_note. Or maybe a custom request handler plugin could actually check that matches are for note_id[x], note_date[x], and note_Text[x] ? Not sure if this is doable. Andre David T. Webb wrote: I have a normalized database schema that I have flattened out to create a Solr schema. My question is with regards to searching the multivalued fields that are correlated from the sub-entity in the DataInputHandler. Example I have 2 tables CUSTOMER and NOTE Customer can have one to many notes. My data-config would look similar to this: (Not exact, just setting up the question) J My schema would be something like this: required="false" multiValued="true" /> All is well, indexed and searchable. So, if there are 100 notes per customer at varying dates, how would I query to essentially ask: Give me all the Customers where note_text has "sales" AND the note_date is between Date1 and Date2? The multi-valued data is stored as arrays and the array positions line up property. (i.e. note_id[x], note_date[x], and note_Text[x] represent an actual row that was loaded from the database. Any suggestions on how to accomplish my problem? Thank you! -- Sincerely, David Webb -- André Bois-Crettez Search technology, Kelkoo http://www.kelkoo.com/
Re: Out of memory during the indexing
How much memory you actually allocate to the JVM ? http://wiki.apache.org/solr/SolrPerformanceFactors#Memory_allocated_to_the_Java_VM You need to increase the -Xmx value, otherwise your large ram buffers won't fit in the java heap. sivaprasad wrote: Hi, I am getting the following error during the indexing.I am trying to index 14 million records but the document size is very minimal. *Error:* 2011-11-08 14:53:24,634 ERROR [STDERR] (Thread-12) java.lang.OutOfMemoryError: GC overhead limit exceeded [...] Do i need to increase the heap size for JVM? My solrconfig settings are given below. false 25 2 1024 2147483647 1 1000 1 and the main index values are false 512 10 2147483647 1 Do i need to increase the ramBufferSizeMB to a little higher? Please provide your inputs. Regards, Siva -- View this message in context: http://lucene.472066.n3.nabble.com/Out-of-memory-during-the-indexing-tp3492701p3492701.html Sent from the Solr - User mailing list archive at Nabble.com. -- André Bois-Crettez Search technology, Kelkoo http://www.kelkoo.com/
Re: Search Correlated Data between Multivalued Fields
Can you point me to the docs on how to create the additional flat index of note? Thx for the quick reply. Dave. Sent from my iPhone On Nov 9, 2011, at 6:03 AM, "Andre Bois-Crettez" wrote: > I do not think this is possbile directly out of the box in Solr. > > A quick workaround would be to fully denormalize the data, ie instead of > multivalued notes for a customer, have a completely flat index of > customer_note. > Or maybe a custom request handler plugin could actually check that matches > are for note_id[x], note_date[x], and note_Text[x] ? Not sure if this is > doable. > > Andre > > David T. Webb wrote: >> I have a normalized database schema that I have flattened out to create >> a Solr schema. My question is with regards to searching the multivalued >> fields that are correlated from the sub-entity in the DataInputHandler. >> >> >> Example >> >> I have 2 tables CUSTOMER and NOTE >> >> >> Customer can have one to many notes. >> >> >> My data-config would look similar to this: (Not exact, just setting up >> the question) J >> >> >> >> >> >> >> >> >> >> My schema would be something like this: >> >> >> > required="true" /> >> >> > required="false" /> >> >> > required="false" /> >> >> > required="false" /> >> >> >> > required="false" multiValued="true" /> >> > required="false" multiValued="true" /> >> >> > required="false" multiValued="true" /> >> >> >> All is well, indexed and searchable. >> >> So, if there are 100 notes per customer at varying dates, how would I >> query to essentially ask: >> >> >> Give me all the Customers where note_text has "sales" AND the note_date >> is between Date1 and Date2? >> >> >> The multi-valued data is stored as arrays and the array positions line >> up property. (i.e. note_id[x], note_date[x], and note_Text[x] represent >> an actual row that was loaded from the database. >> >> >> Any suggestions on how to accomplish my problem? >> >> >> Thank you! >> >> >> -- >> >> Sincerely, >> >> David Webb >> >> >> >> > > -- > André Bois-Crettez > > Search technology, Kelkoo > http://www.kelkoo.com/ >
Solr 4.0 indexing NoSuchMethodError
Hello, I've just installed Solr 4.0, and I am getting an error when indexing. *GRAVE: java.lang.NoSuchMethodError: org.apache.lucene.util.CodecUtil.writeHeader(Lorg/apache/lucene/store/DataOutput;Ljava/lang/String;I)Lorg/apache/lucene/store/DataOutput; at org.apache.lucene.util.fst.FST.save(FST.java:311)*. Does anybody know what I've done wrong? Thanks, Elisabeth
ExtractingRequestHandler HTTP GET Problem
Hi, I've a problem with the ExtractingRequestHandler of Solr. I want to send a really big base64 encoded string to Solr with the CommonsHttpSolrServer. The base64 encoded string is the contet of the indexed file. The CommonsHttpSolrServer sends the parameters as a HTTP GET request. Because of that I'll get a "socket write error". If I change the CommonsHttpSolrServer to send the parameters as HTTP POST sending will work, but the ExtractingRequestHandler will not recognize the parameters. If I'm using the EmbeddedSolrServer there is no problem. This is an option for me now, but I don't know exactly if the solr server will be at another location than the website when my project will be ready for the "real world" use ;-). Is there anything what I can do in the configuration of solr or the application server? Otherwhise I'll write a patch for that, if it'll be accepted. Felix
Re: Search Correlated Data between Multivalued Fields
Something like : David T. Webb wrote: Can you point me to the docs on how to create the additional flat index of note? Thx for the quick reply. Dave. Sent from my iPhone On Nov 9, 2011, at 6:03 AM, "Andre Bois-Crettez" wrote: I do not think this is possbile directly out of the box in Solr. A quick workaround would be to fully denormalize the data, ie instead of multivalued notes for a customer, have a completely flat index of customer_note. Or maybe a custom request handler plugin could actually check that matches are for note_id[x], note_date[x], and note_Text[x] ? Not sure if this is doable. Andre David T. Webb wrote: I have a normalized database schema that I have flattened out to create a Solr schema. My question is with regards to searching the multivalued fields that are correlated from the sub-entity in the DataInputHandler. Example I have 2 tables CUSTOMER and NOTE Customer can have one to many notes. My data-config would look similar to this: (Not exact, just setting up the question) J My schema would be something like this: required="false" multiValued="true" /> required="false" multiValued="true" /> All is well, indexed and searchable. So, if there are 100 notes per customer at varying dates, how would I query to essentially ask: Give me all the Customers where note_text has "sales" AND the note_date is between Date1 and Date2? The multi-valued data is stored as arrays and the array positions line up property. (i.e. note_id[x], note_date[x], and note_Text[x] represent an actual row that was loaded from the database. Any suggestions on how to accomplish my problem? Thank you! -- Sincerely, David Webb -- André Bois-Crettez Search technology, Kelkoo http://www.kelkoo.com/ -- André Bois-Crettez Search technology, Kelkoo http://www.kelkoo.com/
RE: SpellChecker : not getting suggestions for misspelled words
Dali, You might want to try to increase spellcheck.count to something higher, maybe 10 or 20. The default spell checker pre-filters suggestions in such a way that you often need to ask for more results than you actually want to get the right ones. The other thing you might want to see is to go into the "spellchecker" directory after you run your first query and see if it built the dictionary. James Dyer E-Commerce Systems Ingram Content Group (615) 213-4311 -Original Message- From: Dali [mailto:medalibenmans...@gmail.com] Sent: Tuesday, November 08, 2011 5:45 AM To: solr-user@lucene.apache.org Subject: SpellChecker : not getting suggestions for misspelled words Hello everyone ! I'm new to Solr and i'm facing a huge problem since yesterday with the SpellChecker component in Solr. I followed instructions in the wiki page, browsed forums and i don't get any result when typing a misspelled word like other people. Here is what i have: *schema.xml :* ... ... *solrconfig.xml*: explicit velocity browse layout Solritas edismax *:* 10 *,score pr_name^0.5 pr_infGenDesc^1.0 pr_OS^1.2 pr_plus^1.5 pr_techno^10.0 pr_moins^1.1 text,features,name,sku,id,manu,cat 3 pr_name^0.5 pr_infGenDesc^1.0 pr_OS^1.2 pr_plus^1.5 pr_techno^10.0 pr_moins^1.1 on cat manu_exact ipod GB 1 cat,inStock price 0 600 50 after manufacturedate_dt NOW/YEAR-10YEARS NOW +1YEAR before after on text features name 0 name spellcheck ... textSpell default name spellchecker jarowinkler spell org.apache.lucene.search.spell.JaroWinklerDistance spellcheckerJaro true true And here is what i'm typing: http://localhost:8080/solr/select/?q=pr_name:sonadr&spellcheck=true&spellcheck.build=true http://localhost:8080/solr/select/?q=pr_name:sonadr&spellcheck=true&spellcheck.build=true and the correct pr_name value (which is indexed) is "Sonar" Any suggestions ? -- View this message in context: http://lucene.472066.n3.nabble.com/SpellChecker-not-getting-suggestions-for-misspelled-words-tp3490004p3490004.html Sent from the Solr - User mailing list archive at Nabble.com.
RE: SpellChecker : not getting suggestions for misspelled words
You're right James ! It was the solution and i can get suggestions now after increasing spellcheck.count to 20. I also made some change at the URL : http://localhost:8080/solr/spell/?q=pr_name:sonadr&spellcheck=true&spellcheck.build=true instead of: http://localhost:8080/solr/select/?q=pr_name:sonadr&spellcheck=true&spellcheck.build=true Dali -- View this message in context: http://lucene.472066.n3.nabble.com/SpellChecker-not-getting-suggestions-for-misspelled-words-tp3490004p3493762.html Sent from the Solr - User mailing list archive at Nabble.com.
DIH -> how to collect added/error unique keys?
Hi *, I am using DataImportHandler to do imports on a INDEX_QUEUE table (UKEY | ACTION) using a custom Transformer which adds fields from various sources depending on the UKEY. Indexing works fine this way. But now I want to delete the rows from INDEX_QUEUE which were successfully updated. -> Is there a good "API way" to do this? Right now I'm using custom RequestProcessor which collects the UIDs and calls a method on a singleton with access to the DB. It works but I hate these global singletons... :-( public void processAdd(AddUpdateCommand cmd) throws IOException { SolrInputDocument doc = cmd.getSolrInputDocument(); try { super.processAdd(cmd); addOK(doc); } catch (IOException e) { addError(doc); throw e; } catch (RuntimeException e) { addError(doc); throw e; } } Any other suggestions? Regards, Kai Gülzau
Re: Weird: Solr Search result and Analysis Result not match?
Regarding <1>. Take a look at admin/analysis and see the tokenization just to check. Oh, and one more thing... putting in front of kind of defeats the purpose of WordDelimiterFilterFactory. One of the things WDDF does is split on case change and you're removing the case changes before WDDF gets hold of it. Best Erick On Tue, Nov 8, 2011 at 9:40 PM, Ellery Leung wrote: > Thanks Erick, here are my responses: > > 1. Yes. What I want to achieve is that when index is filtered with > EdgeNgram, and a query that is not filtered in that way, I can do search on > partial string. > 2. Good suggestion, will test it. > 3. ok > 4. Thank you > 5/6. Will remove the synonyms and word delimiterfilterfactory in query > 7. will look at that using Luke. By the way, it is the first time I saw that > there is a tool for that. Thank you. > 8. Yes. > > Will check that again, thank you. > > -Original Message- > From: Erick Erickson [mailto:erickerick...@gmail.com] > Sent: 2011年11月8日 9:52 下午 > To: solr-user@lucene.apache.org; elleryle...@be-o.com > Subject: Re: Weird: Solr Search result and Analysis Result not match? > > Several things: > > 1> You don't have EdgeNGramFilterFactory in your query analysis chain, > is this intentional? > 2> You have a LOT of stuff going on here, you might try making your > analysis chain simpler and > adding stuff back in until you see the error. Don't forget to re-index! > 3> Analysis doesn't take into account query *parsing*, so it's > possible to get a false sense of > assurance when the analysis page matches your expectations. > 4> Even though nothing jumps out at me except the Edge factory, > nice job of including > information. > 5> It's unusual to expand synonyms both at query and index time, > usually one or the > other with index time preferred. > 6> Same with WordDelimiterFilterFactory. If you put all the variants > in the index, you don't > need to put all the variants in the query and vice-versa. > 7> Take a look at your actual contents, perhaps using Luke to insure > that what you expect > to be in your index actually is. > 8> You did re-index after your latest changes to your schema, right ? > > All of this is a way of saying that I don't quite see what the problem > is, but at least there are > some avenues to explore. > > Best > Erick > > On Mon, Nov 7, 2011 at 9:29 PM, Ellery Leung wrote: >> Hi all. >> >> >> >> I am using Solr 3.4 under Win 7. >> >> >> >> In schema there is a multivalue field indexed in this way: >> >> == >> >> Schema: >> >> == >> >> > stored="true" omitNorms="true"/> >> >> >> >> > positionIncrementGap="100"> >> >> >> >> > mapping="../../filters/filter-mappings.txt"/> >> >> >> >> >> >> >> >> >> >> > synonyms="../../filters/filter-synonyms.txt" ignoreCase="true" >> expand="true"/> >> >> >> >> > splitOnCaseChange="1" splitOnNumerics="1" stemEnglishPossessive="1" >> generateWordParts="1" generateNumberParts="1" catenateWords="1" >> catenateNumbers="1" catenateAll="0" preserveOriginal="1"/> >> >> > encoder="DoubleMetaphone" inject="true"/> >> >> >> >> > maxGramSize="50" side="front"/> >> >> >> >> >> >> >> >> > mapping="../../filters/filter-mappings.txt"/> >> >> >> >> >> >> >> >> >> >> > synonyms="../../filters/filter-synonyms.txt" ignoreCase="true" >> expand="true"/> >> >> >> >> > splitOnCaseChange="1" splitOnNumerics="1" stemEnglishPossessive="1" >> generateWordParts="0" generateNumberParts="1" catenateWords="1" >> catenateNumbers="1" catenateAll="0" preserveOriginal="1"/> >> >> > encoder="DoubleMetaphone"/> >> >> >> >> >> >> >> >> >> >> == >> >> Actual index: >> >> == >> >> >> >> 2284e2 >> >> 2284e4 >> >> 2284e5 >> >> 1911e2 >> >> >> >> >> >> == >> >> Question: >> >> == >> >> Now when I do a search like this: >> >> >> >> myEvent:1911e2 >> >> >> >> This should match the 4th item. Now on "Full Interface", it does not return >> any result. But on "analysis", matches are highlighted. >> >> >> >> By using Debug: the parsedquery is: >> >> >> >> MultiPhraseQuery(myEvent:"(1911e2 1911) (A e) 2") >> >> >> >> Parsedquery_toString: >> >> >> >> myEvent:"(1911e2 1911) (A e) 2" >> >> >> >> Can anyone please help me on this? >> >> > >
Distributed indexing
Hello! I was looking for a way to implement distributed indexing in Solr. From looking at the https://issues.apache.org/jira/browse/SOLR-2358 there was some work done to enable Solr to distribute the documents to shards without the need of 3rd party software before Solr. What I would like to know if this is the road Solr will take to make things work. Maybe there was some additional work done with distributed indexing ? -- Regards, Rafał Kuć
Re: Solr dismax scoring and weight
Length normalization is an attempt to factor in how long the field is. The idea is that a token in a field with 10,000 tokens should count less than the word in a field of 10 tokens. But since the length of the field is encoded in a byte, the distinction between 4 and 20 characters is pretty much lost. HTH Erick On Wed, Nov 9, 2011 at 3:59 AM, darul wrote: > Thanks for the details, but what do you mean by normalization, can you > describe shortly the concepts behind ? > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Solr-dismax-scoring-and-weight-tp3490096p3492986.html > Sent from the Solr - User mailing list archive at Nabble.com. >
Re: Weird: Solr Search result and Analysis Result not match?
Oh, one more thing. I wasn't suggesting that you *remove* WordDelimiterFilterFactory from the query chain, just that you should be more selective about the options. Look at the differences in the options in the example schema for a place to start Best Erick On Wed, Nov 9, 2011 at 12:33 PM, Erick Erickson wrote: > Regarding <1>. Take a look at admin/analysis and see the tokenization just > to check. > > Oh, and one more thing... > putting in front of > kind of defeats the purpose of WordDelimiterFilterFactory. One of the > things WDDF does is split on case change and you're removing the case > changes before WDDF gets hold of it. > > Best > Erick > > On Tue, Nov 8, 2011 at 9:40 PM, Ellery Leung wrote: >> Thanks Erick, here are my responses: >> >> 1. Yes. What I want to achieve is that when index is filtered with >> EdgeNgram, and a query that is not filtered in that way, I can do search on >> partial string. >> 2. Good suggestion, will test it. >> 3. ok >> 4. Thank you >> 5/6. Will remove the synonyms and word delimiterfilterfactory in query >> 7. will look at that using Luke. By the way, it is the first time I saw >> that there is a tool for that. Thank you. >> 8. Yes. >> >> Will check that again, thank you. >> >> -Original Message- >> From: Erick Erickson [mailto:erickerick...@gmail.com] >> Sent: 2011年11月8日 9:52 下午 >> To: solr-user@lucene.apache.org; elleryle...@be-o.com >> Subject: Re: Weird: Solr Search result and Analysis Result not match? >> >> Several things: >> >> 1> You don't have EdgeNGramFilterFactory in your query analysis chain, >> is this intentional? >> 2> You have a LOT of stuff going on here, you might try making your >> analysis chain simpler and >> adding stuff back in until you see the error. Don't forget to re-index! >> 3> Analysis doesn't take into account query *parsing*, so it's >> possible to get a false sense of >> assurance when the analysis page matches your expectations. >> 4> Even though nothing jumps out at me except the Edge factory, >> nice job of including >> information. >> 5> It's unusual to expand synonyms both at query and index time, >> usually one or the >> other with index time preferred. >> 6> Same with WordDelimiterFilterFactory. If you put all the variants >> in the index, you don't >> need to put all the variants in the query and vice-versa. >> 7> Take a look at your actual contents, perhaps using Luke to insure >> that what you expect >> to be in your index actually is. >> 8> You did re-index after your latest changes to your schema, right ? >> >> All of this is a way of saying that I don't quite see what the problem >> is, but at least there are >> some avenues to explore. >> >> Best >> Erick >> >> On Mon, Nov 7, 2011 at 9:29 PM, Ellery Leung wrote: >>> Hi all. >>> >>> >>> >>> I am using Solr 3.4 under Win 7. >>> >>> >>> >>> In schema there is a multivalue field indexed in this way: >>> >>> == >>> >>> Schema: >>> >>> == >>> >>> >> stored="true" omitNorms="true"/> >>> >>> >>> >>> >> positionIncrementGap="100"> >>> >>> >>> >>> >> mapping="../../filters/filter-mappings.txt"/> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >> synonyms="../../filters/filter-synonyms.txt" ignoreCase="true" >>> expand="true"/> >>> >>> >>> >>> >> splitOnCaseChange="1" splitOnNumerics="1" stemEnglishPossessive="1" >>> generateWordParts="1" generateNumberParts="1" catenateWords="1" >>> catenateNumbers="1" catenateAll="0" preserveOriginal="1"/> >>> >>> >> encoder="DoubleMetaphone" inject="true"/> >>> >>> >>> >>> >> maxGramSize="50" side="front"/> >>> >>> >>> >>> >>> >>> >>> >>> >> mapping="../../filters/filter-mappings.txt"/> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >> synonyms="../../filters/filter-synonyms.txt" ignoreCase="true" >>> expand="true"/> >>> >>> >>> >>> >> splitOnCaseChange="1" splitOnNumerics="1" stemEnglishPossessive="1" >>> generateWordParts="0" generateNumberParts="1" catenateWords="1" >>> catenateNumbers="1" catenateAll="0" preserveOriginal="1"/> >>> >>> >> encoder="DoubleMetaphone"/> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> == >>> >>> Actual index: >>> >>> == >>> >>> >>> >>> 2284e2 >>> >>> 2284e4 >>> >>> 2284e5 >>> >>> 1911e2 >>> >>> >>> >>> >>> >>> == >>> >>> Question: >>> >>> == >>> >>> Now when I do a search like this: >>> >>> >>> >>> myEvent:1911e2 >>> >>> >>> >>> This should match the 4th item. Now on "Full Interface", it does not return >>> any result. But on "analysis", matches are highlig
Importing Big Data From Berkeley DB to Solr
Hi: I have a massive data repository (hundreds of millions of records) stored in Berkeley DB with Java code to access it, and I need an efficient method to import it into Solr for indexing. I cannot find a straightforward Java data import API that I can load the data with. There is no JDBC for the DataImportHandler to call, it is not a simple file, and the inefficiencies (and extra code) needed to submit it as HTTP calls, or as XML feeds, etc. are measures of last resort only. Can a call a Lucene API in a Solr installation to do this somehow? Thanks
Out of memory, not during import or updates of the index
We get at rare times out of memory errors during the day. I know one reason for this is data imports, none are going on. I see in the wiki, document adds have some quirks, not doing that. I don't know to to expect for memory use though. We had Solr running under Tomcat set to 2G ram. I presume cache size has an effect on memory, that's set to 30,000 for filter, document and queryResult. Have experimented with different sizes for a while, these limits are all lower than we used to have them set to. So, hoping there no sort of memory leak involved. In any case, some of the messages are: Exception in thread "http-8080-21" java.lang.OutOfMemoryError: Java heap space Some look like this: Exception in thread "http-8080-22" java.lang.NullPointerException at java.util.concurrent.ConcurrentLinkedQueue.offer(ConcurrentLinkedQueue.java:273) ... I presume the null pointer is a result of being out of memory. Should Solr possibly need more than 2GB? What else can we tune that might reduce memory usage?
Re: Importing Big Data From Berkeley DB to Solr
Carey, Some options: * Just read your BDB and use SolrJ to index to Solr in batches and in parallel * Dump your BDB into csv format and use Solr's ability to import csv files fast * Use Hadoop MapReduce to index to Lucene or Solr in parallel Yes, you can index using Lucene APIs directly, but you will have to make sure all the analysis you specify there is identical to what you have in your Solr schema. Otis Sematext :: http://sematext.com/ :: Solr - Lucene - Nutch Lucene ecosystem search :: http://search-lucene.com/ > >From: Carey Sublette >To: "solr-user@lucene.apache.org" >Sent: Wednesday, November 9, 2011 3:06 PM >Subject: Importing Big Data From Berkeley DB to Solr > >Hi: > >I have a massive data repository (hundreds of millions of records) stored in >Berkeley DB with Java code to access it, and I need an efficient method to >import it into Solr for indexing. I cannot find a straightforward Java data >import API that I can load the data with. > >There is no JDBC for the DataImportHandler to call, it is not a simple file, >and the inefficiencies (and extra code) needed to submit it as HTTP calls, or >as XML feeds, etc. are measures of last resort only. > >Can a call a Lucene API in a Solr installation to do this somehow? > >Thanks > > >
Re: Out of memory, not during import or updates of the index
Hi, Some options: * Yes, on the slave/search side you can reduce your cache sizes and lower the memory footprint. * You can also turn off norms in various fields if you don't need that and save memory there. * You can increase your Xmx I don't know what version of Solr you have, but look through Lucene/Solr's CHANGES.txt to see if there were any changes that affect memory requirements since your version of Solr. Otis Sematext :: http://sematext.com/ :: Solr - Lucene - Nutch Lucene ecosystem search :: http://search-lucene.com/ > >From: Steve Fatula >To: "solr-user@lucene.apache.org" >Sent: Wednesday, November 9, 2011 3:33 PM >Subject: Out of memory, not during import or updates of the index > >We get at rare times out of memory errors during the day. I know one reason >for this is data imports, none are going on. I see in the wiki, document adds >have some quirks, not doing that. I don't know to to expect for memory use >though. > >We had Solr running under Tomcat set to 2G ram. I presume cache size has an >effect on memory, that's set to 30,000 for filter, document and queryResult. >Have experimented with different sizes for a while, these limits are all lower >than we used to have them set to. So, hoping there no sort of memory leak >involved. > >In any case, some of the messages are: > >Exception in thread "http-8080-21" java.lang.OutOfMemoryError: Java heap space > > >Some look like this: > >Exception in thread "http-8080-22" java.lang.NullPointerException > at >java.util.concurrent.ConcurrentLinkedQueue.offer(ConcurrentLinkedQueue.java:273) >... > >I presume the null pointer is a result of being out of memory. > >Should Solr possibly need more than 2GB? What else can we tune that might >reduce memory usage? > >
Re: Solr 4.0 indexing NoSuchMethodError
The CodecUtil.writeHeader signature has changed from public static DataOutput writeHeader(DataOutput out, String codec, int version) in lucene 3.4 (which is the method not found) to public static void writeHeader(DataOutput out, String codec, int version) in lucene 4.0 It means that while you're using solr 4.0, some 3.4 jars are stuck somewhere in the java classpath. Obviously some code is looking for this 3.4 method. If you're using the start.jar executable, you should have a look at your system-wide classpath If you're using tomcat (and that sounds plausible in this situation), you should trash the "work" directory sub-folders of your tomcat installation; and restart it. Tomcat unpacks war archives in this directory, and it may have kept a 3.4 solr war deployed here. 2011/11/9 elisabeth benoit > Hello, > > I've just installed Solr 4.0, and I am getting an error when indexing. > > *GRAVE: java.lang.NoSuchMethodError: > > org.apache.lucene.util.CodecUtil.writeHeader(Lorg/apache/lucene/store/DataOutput;Ljava/lang/String;I)Lorg/apache/lucene/store/DataOutput; >at org.apache.lucene.util.fst.FST.save(FST.java:311)*. > > Does anybody know what I've done wrong? > > Thanks, > Elisabeth >
Re: Out of memory, not during import or updates of the index
From: Otis Gospodnetic >To: "solr-user@lucene.apache.org" >Sent: Wednesday, November 9, 2011 2:51 PM >Subject: Re: Out of memory, not during import or updates of the index > >Hi, > >Some options: >* Yes, on the slave/search side you can reduce your cache sizes and lower the >memory footprint. >* You can also turn off norms in various fields if you don't need that and >save memory there. >* You can increase your Xmx > >I don't know what version of Solr you have, but look through Lucene/Solr's >CHANGES.txt to see if there were any changes that affect memory requirements >since your version of Solr. > > > > > >Using Solr 3.4.0. That changelog actually says it should reduce memory usage >for that version. We were on a much older version previously, 1.something. Norms are off on all fields that it can be turned off on. I'm just hoping this new version doesn't have any leaks. Does FastLRUCache vs LRUCache make any memory difference?
Anyway to stop an optimize?
Hi, Does anyone know if an optimize can be stopped once started? Thanks
RE: Importing Big Data From Berkeley DB to Solr
Thanks Otis: It looks like SolrJ is what I was looking for exactly, it is also nice to know that the csv implementation is fast as a fall back. -Original Message- From: Otis Gospodnetic [mailto:otis_gospodne...@yahoo.com] Sent: Wednesday, November 09, 2011 12:48 PM To: solr-user@lucene.apache.org Subject: Re: Importing Big Data From Berkeley DB to Solr Carey, Some options: * Just read your BDB and use SolrJ to index to Solr in batches and in parallel * Dump your BDB into csv format and use Solr's ability to import csv files fast * Use Hadoop MapReduce to index to Lucene or Solr in parallel Yes, you can index using Lucene APIs directly, but you will have to make sure all the analysis you specify there is identical to what you have in your Solr schema. Otis Sematext :: http://sematext.com/ :: Solr - Lucene - Nutch Lucene ecosystem search :: http://search-lucene.com/ > >From: Carey Sublette >To: "solr-user@lucene.apache.org" >Sent: Wednesday, November 9, 2011 3:06 PM >Subject: Importing Big Data From Berkeley DB to Solr > >Hi: > >I have a massive data repository (hundreds of millions of records) stored in >Berkeley DB with Java code to access it, and I need an efficient method to >import it into Solr for indexing. I cannot find a straightforward Java data >import API that I can load the data with. > >There is no JDBC for the DataImportHandler to call, it is not a simple file, >and the inefficiencies (and extra code) needed to submit it as HTTP calls, or >as XML feeds, etc. are measures of last resort only. > >Can a call a Lucene API in a Solr installation to do this somehow? > >Thanks > > >
Unable to determine why query won't return results
Hello all. I'm having an issue in regards to matching a quoted phrase in Solr, and I'm not certain what the issue at hand is. I have tried this on both Solr 1.3 (Our production system) and 3.3 (Our development system). The field is a text field, and has the following fieldType definition: http://pastebin.com/SkmmucUE In the case where the search is failing, the field is indexed with the following value: A. J. Johnson & Co. We are searching the field with the following string (in quotes): "A. J. Johnson & Co." Unfortunately, we get a response of no results when searching the field in question with the above specified string. If we search merely for "A. J. Johnson" (with quotes), we get the desired result. Using the full string, however, seems to cause the results not to match. I have attempted to use Solr's analyzer (without success) to trace the problem. The results of this are here: http://pastehtml.com/view/bdgpdrt0w.html Any suggestions?
Re: Anyway to stop an optimize?
Don't think so, at least not gracefully. You can always do partial optimize and do a few of them if you want to optimize in smaller steps. Otis > >From: Brendan Grainger >To: solr-user@lucene.apache.org >Sent: Wednesday, November 9, 2011 4:35 PM >Subject: Anyway to stop an optimize? > >Hi, > >Does anyone know if an optimize can be stopped once started? > >Thanks > > >
Re: Anyway to stop an optimize?
If you restart the server, the optimize should stop and not restart, right? wunder On Nov 9, 2011, at 7:43 PM, Otis Gospodnetic wrote: > Don't think so, at least not gracefully. You can always do partial optimize > and do a few of them if you want to optimize in smaller steps. > > Otis > > >> >> From: Brendan Grainger >> To: solr-user@lucene.apache.org >> Sent: Wednesday, November 9, 2011 4:35 PM >> Subject: Anyway to stop an optimize? >> >> Hi, >> >> Does anyone know if an optimize can be stopped once started? >> >> Thanks >> >>
Re: Anyway to stop an optimize?
I think in the past I've tried that, and it has restarted, although I will have to try it out (this time we were loath to stop it as we didn't want any index corruption issues). A related question is, why did the optimize start? I thought it had to be explicitly started, but somehow it started optimzing on it's own. Thanks again Brendan On Nov 9, 2011, at 10:44 PM, Walter Underwood wrote: > If you restart the server, the optimize should stop and not restart, right? > > wunder > > On Nov 9, 2011, at 7:43 PM, Otis Gospodnetic wrote: > >> Don't think so, at least not gracefully. You can always do partial optimize >> and do a few of them if you want to optimize in smaller steps. >> >> Otis >> >> >>> >>> From: Brendan Grainger >>> To: solr-user@lucene.apache.org >>> Sent: Wednesday, November 9, 2011 4:35 PM >>> Subject: Anyway to stop an optimize? >>> >>> Hi, >>> >>> Does anyone know if an optimize can be stopped once started? >>> >>> Thanks >>> >>> > > > >
Re: Anyway to stop an optimize?
A restart during an optimize should not cause index corruption. The optimize only reads existing indexes, and the only writes are to indexes not yet in use. If it does not finish, those half-written indexes are junk to be cleaned up later. wunder On Nov 9, 2011, at 8:16 PM, Brendan Grainger wrote: > I think in the past I've tried that, and it has restarted, although I will > have to try it out (this time we were loath to stop it as we didn't want any > index corruption issues). > > A related question is, why did the optimize start? I thought it had to be > explicitly started, but somehow it started optimzing on it's own. > > Thanks again > Brendan > > On Nov 9, 2011, at 10:44 PM, Walter Underwood wrote: > >> If you restart the server, the optimize should stop and not restart, right? >> >> wunder >> >> On Nov 9, 2011, at 7:43 PM, Otis Gospodnetic wrote: >> >>> Don't think so, at least not gracefully. You can always do partial >>> optimize and do a few of them if you want to optimize in smaller steps. >>> >>> Otis >>> >>> From: Brendan Grainger To: solr-user@lucene.apache.org Sent: Wednesday, November 9, 2011 4:35 PM Subject: Anyway to stop an optimize? Hi, Does anyone know if an optimize can be stopped once started? Thanks >>
Re: Anyway to stop an optimize?
This is correct. And there is no way I can think of optimize could just start on its own - somebody or something called it. Otis Sematext :: http://sematext.com/ :: Solr - Lucene - Nutch Lucene ecosystem search :: http://search-lucene.com/ > >From: Walter Underwood >To: solr-user@lucene.apache.org >Sent: Wednesday, November 9, 2011 11:38 PM >Subject: Re: Anyway to stop an optimize? > >A restart during an optimize should not cause index corruption. The optimize >only reads existing indexes, and the only writes are to indexes not yet in >use. If it does not finish, those half-written indexes are junk to be cleaned >up later. > >wunder > >On Nov 9, 2011, at 8:16 PM, Brendan Grainger wrote: > >> I think in the past I've tried that, and it has restarted, although I will >> have to try it out (this time we were loath to stop it as we didn't want any >> index corruption issues). >> >> A related question is, why did the optimize start? I thought it had to be >> explicitly started, but somehow it started optimzing on it's own. >> >> Thanks again >> Brendan >> >> On Nov 9, 2011, at 10:44 PM, Walter Underwood wrote: >> >>> If you restart the server, the optimize should stop and not restart, right? >>> >>> wunder >>> >>> On Nov 9, 2011, at 7:43 PM, Otis Gospodnetic wrote: >>> Don't think so, at least not gracefully. You can always do partial optimize and do a few of them if you want to optimize in smaller steps. Otis > > From: Brendan Grainger > To: solr-user@lucene.apache.org > Sent: Wednesday, November 9, 2011 4:35 PM > Subject: Anyway to stop an optimize? > > Hi, > > Does anyone know if an optimize can be stopped once started? > > Thanks > > >>> > > > > > > >
Dynamic adding of shards
Hi, One way to add new shards is to add them in the shard parameter in the solrconfig.xml file. But this will require to restart the solr server everytime you add a new shard. I wanted to know if it is possible to dynamically add shards without having to restart the solr server. If yes how? Thanks in advance. Ankita
abort processing query
Hi all We have very complexed queries including wildcard. That causes memory overhead. Sometimes, memory is full and server doesn't response. What I wonder, when query process time on server exceeds the time limit, can I abort processing query? If possible, how should I do? Thanks in advance Jason -- View this message in context: http://lucene.472066.n3.nabble.com/abort-processing-query-tp3495876p3495876.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: Aggregated indexing of updating RSS feeds
All, Can anyone advise how to stop the "deleteAll" event during a full import? I'm still unable to determine why repeat full imports seem to delete old indexes. After investigation the logs confirm this - see "REMOVING ALL DOCUMENTS FROM INDEX" below. ..but the request I'm making is.. /solr/myfeed?command=full-import&rows=5000&clean=false ..note the clean=false. All help appreciated. Shaun INFO: [] webapp=/solr path=/myfeed params={command=full-import} status=0 QTime=8 10-Nov-2011 05:40:01 org.apache.solr.handler.dataimport.DataImporter doFullImport INFO: Starting Full Import 10-Nov-2011 05:40:01 org.apache.solr.handler.dataimport.SolrWriter readIndexerProperties INFO: Read myfeed.properties 10-Nov-2011 05:40:01 org.apache.solr.update.DirectUpdateHandler2 deleteAll INFO: [] REMOVING ALL DOCUMENTS FROM INDEX 10-Nov-2011 05:40:05 org.apache.solr.core.SolrCore execute INFO: [] webapp=/solr path=/select/ params={indent=on&start=0&q=description:one+direction&rows=10&version=2.2} hits=0 status=0 QTime=1 10-Nov-2011 05:40:07 org.apache.solr.core.SolrCore execute INFO: [] webapp=/solr path=/select/ params={indent=on&start=0&q=id:*23327977*&rows=10&version=2.2} hits=0 status=0 QTime=1 10-Nov-2011 05:40:08 org.apache.solr.core.SolrDeletionPolicy onInit INFO: SolrDeletionPolicy.onInit: commits:num=2000 commit{dir=/mnt/ebs1/data/index,segFN=segments_1x3,version=1319402557686,generation=2487,filenames=[_3u3.tii, segments_1x3, _3u3.frq, _3u3.prx, _3u3.nrm, _3u3.fnm, _3u3.fdx, _3u3.tis, _3u3.fdt] commit{dir=/mnt/ebs1/data/index,segFN=segments_1x4,version=1319402557691,generation=2488,filenames=[_3u5.nrm, _3u5.fnm, _3u5.fdx, segments_1x4, _3u5.tis, _3u5.prx, _3u5.frq, _3u5.tii, _3u5.fdt] -- View this message in context: http://lucene.472066.n3.nabble.com/Aggregated-indexing-of-updating-RSS-feeds-tp3485335p3495882.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: abort processing query
> Hi all > We have very complexed queries including wildcard. > That causes memory overhead. > Sometimes, memory is full and server doesn't response. > What I wonder, when query process time on server exceeds > the time limit, can > I abort processing query? > If possible, how should I do? QueryComponent respects timeAllowed parameter. http://wiki.apache.org/solr/CommonQueryParameters#timeAllowed
Re: About the total size of all files
On Thu, Nov 10, 2011 at 12:39 PM, 刘浪 wrote: > > Hi, > The total size of all files can attach PB or TB? > If I use the only one solr core to index PB level files, how about the > time of searching? Can the time will be less 1 second? > If I use the only multi solr core to index PB level files, how about the > time of searching? Can the time will be less 1 second? It is difficult to understand your question. What is PB, and TB? Peta-byte/Tera-byte? It is difficult to make predictions about indexing, and search times without understanding the volume and complexity of your data. The best way to get an estimate of this is to set up Solr, and try this out, maybe on a subset of the data at first. Regards, Gora