lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Pranaya Behera <pranaya.beh...@igp.com>
Subject Re: Inconsistent results with solr admin ui and solrj
Date Tue, 16 Aug 2016 04:51:50 GMT
Hi,
  a.) Yes index is static, not updated live. We index new documents over 
old documents by this sequesce, deleteall docs, add 10 freshly fetched 
from db, after adding all the docs to cloud instance, commit. Commit 
happens only once per collection,
b.) I took one shard and below are the results for the each replica, it 
has 2 replica.
Replica - 2
Last Modified: 33 minutes ago
Num Docs: 127970
Max Doc: 127970
Heap Memory Usage: -1
Deleted Docs: 0
Version: 14530
Segment Count: 5
Optimized: yes
Current: yes
Data:  /var/solr/data/product_shard1_replica2/data
Index: /var/solr/data/product_shard1_replica2/data/index.20160816040537452
Impl:  org.apache.solr.core.NRTCachingDirectoryFactory

Replica - 1
Last Modified: about 19 hours ago
Num Docs: 234013
Max Doc: 234013
Heap Memory Usage: -1
Deleted Docs: 0
Version: 14272
Segment Count: 7
Optimized: yes
Current: no
Data:  /var/solr/data/product_shard1_replica1/data
Index: /var/solr/data/product_shard1_replica1/data/index
Impl:  org.apache.solr.core.NRTCachingDirectoryFactory

c.) With the admin ui: if I query for all, *:* it gives different 
numFound each time.
e.g.
1.

|{ "responseHeader":{ "zkConnected":true, "status":0, "QTime":7, 
"params":{ "q":"*:*", "indent":"on", "wt":"json", "_":"1471322871767"}}, 
"response":{"numFound":452300,"start":0,"maxScore":1.0, 2. |
|{ "responseHeader":{ "zkConnected":true, "status":0, "QTime":23, 
"params":{ "q":"*:*", "indent":"on", "wt":"json", "_":"1471322871767"}}, 
"response":{"numFound":574013,"start":0,"maxScore":1.0, This is queried 
live from the solr instances. |

It happens with any type of queries, if I search in parent document or 
search through child documents to get parents. sorting is used in both 
cases but with different field, while doingblock join query sortingis on 
the child document field, otherwise on the parent document field.

d.) I dont find any errors in the logs. All warnings only.

On 14/08/16 02:56, Jan Høydahl wrote:
> Could it be that your cluster is not in sync, so that when Solr picks three nodes, results
will vary depending on what replica answers?
>
> A few questions:
>
> a) Is your index static, i.e. not being updated live?
> b) Can you try to go directly to the core menu of both replicas for each shard, and compare
numDocs / maxDocs for each? Both replicas in each shard should have same count.
> c) What are you querying on and sorting by? Does it happen with only one query and sorting?
> d) Are there any errors in the logs?
>
> If possible, please share some queries, responses, config, screenshots etc.
>
> --
> Jan Høydahl, search solution architect
> Cominvent AS - www.cominvent.com
>
>> 13. aug. 2016 kl. 12.10 skrev Pranaya Behera <pranaya.behera@igp.com>:
>>
>> Hi,
>>     I am running solr 6.1.0 with solrcloud. We have 3 instance of zookeeper and 3
instance of solrcloud. All three of them are active and up. One collection has 3 shards, each
shard has 2 replicas.
>>
>> Everytime query whether from solrj or admin ui, getting inconsistent results. e.g.
>> 1. numFound is always fluctuating.
>> 2. facet count shows the count for a field, filter query on that field gets 0 results.
>> 3. luke requests work(not sure whether gives correct info of all the dynamic field)
on per shard not on collection when invoked from curl but doesnt work when called from solrj.
>> 4. admin ui shows expanded results, same query goes from solrj, getExpandedResults()
gives 0 docs.
>>
>> What would be cause of all this ? Any pointer to look for an error anything in the
logs.
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message