[jira] [Created] (SOLR-2508) Master/Slave replication can leave slave in inconsistent state of NullPointerException in solrHighligher.java 102

2011-05-11 Thread Xing Li (JIRA) 
Master/Slave replication can leave slave in inconsistent state of  
NullPointerException in solrHighligher.java 102
--

 Key: SOLR-2508
 URL: https://issues.apache.org/jira/browse/SOLR-2508
 Project: Solr
  Issue Type: Bug
  Components: highlighter, replication (java)
Affects Versions: 4.0
 Environment: Centos 5.6 with Java1.7.0b137
Reporter: Xing Li


Using Solr 4/Trunk snapshot build of 5/10/2011. 

Setup:
--
1) 1 Master + 4 Slaves
2) Multicore setup with 8 cores.
3) Replication Poll Interval: 00:30:20

Summary of Issue:
---
When a slave completes a replication pull from master, it will complete the 
data index pull but 
based on logs it appears subsequent index warming and other actions post 
replication 
cleanup leaves the core/db in an inconsistent state.

Frequency of occurrence: Very high but not 100%. I have 1 master and 4 slaves 
and for each replication 
pull cycle, around 50% of the gets affected. Each slave has 8 multi-cores but
the problem always affects this particular mysolr_blogs db/core.

Please note the mysolr_blogs data index is 1.4GB and the largest of the 8 by 
a wide margin.

Attached is the schema.xml and solrconfig.xml for the mysolr_blogs core.


Temp fix:
-
1) Stop and restart the solr server when this happens.
2) Stop using automatic replication on this core.


Logging:
-

* begins automatic replication  pull

{code}
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave in sync with master.
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave in sync with master.
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave in sync with master.
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave in sync with master.
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave in sync with master.
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave in sync with master.
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Master's version: 1302675975227, generation: 694
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave in sync with master.
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Slave's version: 1302675975222, generation: 692
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Starting replication process
May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Number of files in latest index in master: 10
{code}

* 65 seconds past and I cut out the query logs in between. Here it's pulling 
the 1.4GB mysolr_blogs index data. 

{code}
May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller downloadIndexFiles
INFO: Skipping download for 
/db/solr-master/multicore/mysolr_blogs/data/index/1.fnx
May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
INFO: Total time taken for download : 65 secs
May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
INFO: [mysolr_users] webapp=/solr path=/select 
params={sort=indent=offstart=0q=%2Buname:inlove*q.op=andhl.fl=*facet.field=pcategoryidfacet.field=categoryidfacet.field=languageidwt=jsonhl=truerows=51}
 hits=0 status=0 QTime=1 
May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 status=0 
QTime=0 
May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 status=0 
QTime=0 
May 10, 2011 10:18:46 PM org.apache.solr.update.DirectUpdateHandler2 commit
INFO: start 
commit(optimize=false,waitFlush=true,waitSearcher=true,expungeDeletes=false)
May 10, 2011 10:18:46 PM org.apache.solr.search.SolrIndexSearcher init
INFO: Opening Searcher@4f83f9df main
May 10, 2011 10:18:46 PM org.apache.solr.update.DirectUpdateHandler2 commit
INFO: end_commit_flush
May 10, 2011 10:18:46 PM org.apache.solr.search.SolrIndexSearcher warm
INFO: autowarming Searcher@4f83f9df main from Searcher@5f7808af main

[jira] [Updated] (SOLR-2508) Master/Slave replication can leave slave in inconsistent state of NullPointerException in solrHighligher.java 102

2011-05-11 Thread Xing Li (JIRA) 

 [ 
https://issues.apache.org/jira/browse/SOLR-2508?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Xing Li updated SOLR-2508:
--

Attachment: (was: schema.xml)

 Master/Slave replication can leave slave in inconsistent state of  
 NullPointerException in solrHighligher.java 102
 --

 Key: SOLR-2508
 URL: https://issues.apache.org/jira/browse/SOLR-2508
 Project: Solr
  Issue Type: Bug
  Components: highlighter, replication (java)
Affects Versions: 4.0
 Environment: Centos 5.6 with Java1.7.0b137
Reporter: Xing Li

 Using Solr 4/Trunk snapshot build of 5/10/2011. 
 Setup:
 --
 1) 1 Master + 4 Slaves
 2) Multicore setup with 8 cores.
 3) Replication Poll Interval: 00:30:20
 Summary of Issue:
 ---
 When a slave completes a replication pull from master, it will complete the 
 data index pull but 
 based on logs it appears subsequent index warming and other actions post 
 replication 
 cleanup leaves the core/db in an inconsistent state.
 Frequency of occurrence: Very high but not 100%. I have 1 master and 4 slaves 
 and for each replication 
 pull cycle, around 50% of the gets affected. Each slave has 8 multi-cores but
 the problem always affects this particular mysolr_blogs db/core.
 Please note the mysolr_blogs data index is 1.4GB and the largest of the 8 
 by a wide margin.
 Attached is the schema.xml and solrconfig.xml for the mysolr_blogs core.
 Temp fix:
 -
 1) Stop and restart the solr server when this happens.
 2) Stop using automatic replication on this core.
 Logging:
 -
 * begins automatic replication  pull
 {code}
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Master's version: 1302675975227, generation: 694
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave's version: 1302675975222, generation: 692
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Starting replication process
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Number of files in latest index in master: 10
 {code}
 * 65 seconds past and I cut out the query logs in between. Here it's pulling 
 the 1.4GB mysolr_blogs index data. 
 {code}
 May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller downloadIndexFiles
 INFO: Skipping download for 
 /db/solr-master/multicore/mysolr_blogs/data/index/1.fnx
 May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Total time taken for download : 65 secs
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_users] webapp=/solr path=/select 
 params={sort=indent=offstart=0q=%2Buname:inlove*q.op=andhl.fl=*facet.field=pcategoryidfacet.field=categoryidfacet.field=languageidwt=jsonhl=truerows=51}
  hits=0 status=0 QTime=1 
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 
 status=0 QTime=0 
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 
 status=0 QTime=0 
 May 10, 2011 10:18:46 PM org.apache.solr.update.DirectUpdateHandler2 commit
 INFO: start 
 commit(optimize=false,waitFlush=true,waitSearcher=true,expungeDeletes=false)
 May 10, 2011 10:18:46 PM org.apache.solr.search.SolrIndexSearcher init
 INFO: Opening Searcher@4f83f9df main
 May 10, 2011 10:18:46 PM org.apache.solr.update.DirectUpdateHandler2 commit
 INFO: end_commit_flush
 May 10, 2011 10:18:46 PM org.apache.solr.search.SolrIndexSearcher warm
 INFO: autowarming Searcher@4f83f9df main from Searcher@5f7808af main

[jira] [Updated] (SOLR-2508) Master/Slave replication can leave slave in inconsistent state of NullPointerException in solrHighligher.java 102

2011-05-11 Thread Xing Li (JIRA) 

 [ 
https://issues.apache.org/jira/browse/SOLR-2508?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Xing Li updated SOLR-2508:
--

Attachment: solrconfig.xml
schema.xml

 Master/Slave replication can leave slave in inconsistent state of  
 NullPointerException in solrHighligher.java 102
 --

 Key: SOLR-2508
 URL: https://issues.apache.org/jira/browse/SOLR-2508
 Project: Solr
  Issue Type: Bug
  Components: highlighter, replication (java)
Affects Versions: 4.0
 Environment: Centos 5.6 with Java1.7.0b137
Reporter: Xing Li

 Using Solr 4/Trunk snapshot build of 5/10/2011. 
 Setup:
 --
 1) 1 Master + 4 Slaves
 2) Multicore setup with 8 cores.
 3) Replication Poll Interval: 00:30:20
 Summary of Issue:
 ---
 When a slave completes a replication pull from master, it will complete the 
 data index pull but 
 based on logs it appears subsequent index warming and other actions post 
 replication 
 cleanup leaves the core/db in an inconsistent state.
 Frequency of occurrence: Very high but not 100%. I have 1 master and 4 slaves 
 and for each replication 
 pull cycle, around 50% of the gets affected. Each slave has 8 multi-cores but
 the problem always affects this particular mysolr_blogs db/core.
 Please note the mysolr_blogs data index is 1.4GB and the largest of the 8 
 by a wide margin.
 Attached is the schema.xml and solrconfig.xml for the mysolr_blogs core.
 Temp fix:
 -
 1) Stop and restart the solr server when this happens.
 2) Stop using automatic replication on this core.
 Logging:
 -
 * begins automatic replication  pull
 {code}
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Master's version: 1302675975227, generation: 694
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave's version: 1302675975222, generation: 692
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Starting replication process
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Number of files in latest index in master: 10
 {code}
 * 65 seconds past and I cut out the query logs in between. Here it's pulling 
 the 1.4GB mysolr_blogs index data. 
 {code}
 May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller downloadIndexFiles
 INFO: Skipping download for 
 /db/solr-master/multicore/mysolr_blogs/data/index/1.fnx
 May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Total time taken for download : 65 secs
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_users] webapp=/solr path=/select 
 params={sort=indent=offstart=0q=%2Buname:inlove*q.op=andhl.fl=*facet.field=pcategoryidfacet.field=categoryidfacet.field=languageidwt=jsonhl=truerows=51}
  hits=0 status=0 QTime=1 
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 
 status=0 QTime=0 
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 
 status=0 QTime=0 
 May 10, 2011 10:18:46 PM org.apache.solr.update.DirectUpdateHandler2 commit
 INFO: start 
 commit(optimize=false,waitFlush=true,waitSearcher=true,expungeDeletes=false)
 May 10, 2011 10:18:46 PM org.apache.solr.search.SolrIndexSearcher init
 INFO: Opening Searcher@4f83f9df main
 May 10, 2011 10:18:46 PM org.apache.solr.update.DirectUpdateHandler2 commit
 INFO: end_commit_flush
 May 10, 2011 10:18:46 PM org.apache.solr.search.SolrIndexSearcher warm
 INFO: autowarming Searcher@4f83f9df main from Searcher@5f7808af main

[jira] [Commented] (SOLR-2508) Master/Slave replication can leave slave in inconsistent state of NullPointerException in solrHighligher.java 102

2011-05-11 Thread Xing Li (JIRA) 

[ 
https://issues.apache.org/jira/browse/SOLR-2508?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13031590#comment-13031590
 ] 

Xing Li commented on SOLR-2508:
---

Got the problem more isolated. 

The queries effected are those using hl.fl=* with hl=true. 

All queries run fine until something in the post replication triggers failures 
of NullPointerException in solrHighligher.java 102 when wildcard is used for 
highlighting field selection hl.fl=*. Replacing those queries with a specific 
highlight field such as hl.fl=uname will then make the query work post 
sudden failure. 

 

 Master/Slave replication can leave slave in inconsistent state of  
 NullPointerException in solrHighligher.java 102
 --

 Key: SOLR-2508
 URL: https://issues.apache.org/jira/browse/SOLR-2508
 Project: Solr
  Issue Type: Bug
  Components: highlighter, replication (java)
Affects Versions: 4.0
 Environment: Centos 5.6 with Java1.7.0b137
Reporter: Xing Li
 Attachments: schema.xml, solrconfig.xml


 Using Solr 4/Trunk snapshot build of 5/10/2011. 
 Setup:
 --
 1) 1 Master + 4 Slaves
 2) Multicore setup with 8 cores.
 3) Replication Poll Interval: 00:30:20
 Summary of Issue:
 ---
 When a slave completes a replication pull from master, it will complete the 
 data index pull but 
 based on logs it appears subsequent index warming and other actions post 
 replication 
 cleanup leaves the core/db in an inconsistent state.
 Frequency of occurrence: Very high but not 100%. I have 1 master and 4 slaves 
 and for each replication 
 pull cycle, around 50% of the gets affected. Each slave has 8 multi-cores but
 the problem always affects this particular mysolr_blogs db/core.
 Please note the mysolr_blogs data index is 1.4GB and the largest of the 8 
 by a wide margin.
 Attached is the schema.xml and solrconfig.xml for the mysolr_blogs core.
 Temp fix:
 -
 1) Stop and restart the solr server when this happens.
 2) Stop using automatic replication on this core.
 Logging:
 -
 * begins automatic replication  pull
 {code}
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Master's version: 1302675975227, generation: 694
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave in sync with master.
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Slave's version: 1302675975222, generation: 692
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Starting replication process
 May 10, 2011 10:17:40 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Number of files in latest index in master: 10
 {code}
 * 65 seconds past and I cut out the query logs in between. Here it's pulling 
 the 1.4GB mysolr_blogs index data. 
 {code}
 May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller downloadIndexFiles
 INFO: Skipping download for 
 /db/solr-master/multicore/mysolr_blogs/data/index/1.fnx
 May 10, 2011 10:18:45 PM org.apache.solr.handler.SnapPuller fetchLatestIndex
 INFO: Total time taken for download : 65 secs
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_users] webapp=/solr path=/select 
 params={sort=indent=offstart=0q=%2Buname:inlove*q.op=andhl.fl=*facet.field=pcategoryidfacet.field=categoryidfacet.field=languageidwt=jsonhl=truerows=51}
  hits=0 status=0 QTime=1 
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 
 status=0 QTime=0 
 May 10, 2011 10:18:45 PM org.apache.solr.core.SolrCore execute
 INFO: [mysolr_blogs] webapp=/solr path=/select/ params={q=solr} hits=0 
 status=0 QTime=0 
 May 10, 2011 10:18:46 PM org.apache.solr.update.DirectUpdateHandler2 commit
 INFO: start 
 commit(optimize=false,waitFlush=true,waitSearcher=true,expungeDeletes=false)
 May 10, 2011 10:18:46 PM org.apache.solr.search.SolrIndexSearcher init
 INFO: Opening Searcher@4f83f9df main
 May 10, 2011 10:18:46 PM

[jira] Commented: (SOLR-2432) using sort option only returns numFound=1 even if there are many more matching

2011-03-18 Thread Xing Li (JIRA) 

[ 
https://issues.apache.org/jira/browse/SOLR-2432?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13008344#comment-13008344
 ] 

Xing Li commented on SOLR-2432:
---

Once the solr.war is copied to webapps, star the app with

java -Xmx4000M -Dsolr.solr.home=multicore 
-Djava.util.logging.config.file=logging.properties -jar start.jar

The solr ip port will be bound to 11101 and use the test urls in the previous 
post.





 using sort option only returns numFound=1 even if there are many more matching
 --

 Key: SOLR-2432
 URL: https://issues.apache.org/jira/browse/SOLR-2432
 Project: Solr
  Issue Type: Bug
  Components: search
Affects Versions: 4.0
Reporter: Xing Li
  Labels: 4.0, sort, trunk
 Attachments: solr.tgz


 Using 4.0 trunk version as of 3/16/2011
 Not sure at which point in the 4.0 trunk did this problem happen but this did 
 not happen in previous versions.
 Symptom:
 1) Assume you have a search db with a query for love that returns 500 
 results.
 2) Append sort=anyfield desc to the search query and you now only get 1 
 result back.
 I even rebuilt the db from scratch to make sure it was not a corrupted solr 
 db but that is not the case.
 Temporary Fix:
 1) Append ,score desc to the sort value such as sort=anyfield 
 desc,score desc fixes the issue and gives back the correct result.
  

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

[jira] Created: (SOLR-2432) using sort option only returns numFound=1 even if there are many more matching

2011-03-17 Thread Xing Li (JIRA) 
using sort option only returns numFound=1 even if there are many more matching
--

 Key: SOLR-2432
 URL: https://issues.apache.org/jira/browse/SOLR-2432
 Project: Solr
  Issue Type: Bug
  Components: search
Affects Versions: 4.0
Reporter: Xing Li


Using 4.0 trunk version as of 3/16/2011

Not sure at which point in the 4.0 trunk did this problem happen but this did 
not happen in previous versions.

Symptom:

1) Assume you have a search db with a query for love that returns 500 results.
2) Append sort=anyfield desc to the search query and you now only get 1 
result back.

I even rebuilt the db from scratch to make sure it was not a corrupted solr db 
but that is not the case.

Temporary Fix:

1) Append ,score desc to the sort value such as sort=anyfield desc,score 
desc fixes the issue and gives back the correct result.

 


--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org