Solr 5.5 Solr Exception Error opening new searcher (NiFi Flow)

[#1]

Hi - Solr 5.5 was working fine with NiFi, until I restarted my VM. Now when I start Nifi and Solr, I get the following message in Solr:

tweets_shard1_replica1: org.apache.solr.common.SolrException:org.apache.solr.common.SolrException: Error opening new searcher

I'm attaching the complete solr log. Please can someone help?

Thank Youtweets-shard1-replica1.txt

[#2]

The attached file doesn't look like a normal Solr log, did that come from solr_home/server/logs/solr.log ?

[#3]

@Bryan Bende I got the info from Solr UI, under 'Logging'.

[#4]

@Bryan Bende any ideas?

[#5]

<response>

<lst name="responseHeader">

<int name="status">0</int> <int name="QTime">0</int> </lst>

<arr name="collections">

<str>collection1</str> <str>tweets</str> </arr>

</response>

[root@sandbox ~]# su solr

[solr@sandbox root]$ /opt/lucidworks-hdpsearch/solr/bin/solr create -c tweets -d tweet_configs -s 1 -rf 1 -p 8983

Connecting to ZooKeeper at sandbox.hortonworks.com:2181/solr ...

Re-using existing configuration directory tweets

ERROR:

Collection 'tweets' already exists!

Checked collection existence using Collections API command:

http://sandbox.hortonworks.com:8983/solr/admin/collections?action=list

[#6]

@Raf Mohammed It looks like your transaction log has some issue and is preventing solr from opening a new searcher. Have you tried to restart solr or issue a manual commit ? Is your HDFS healthy?

You could remove the recent transaction logs, which should resolve this issue, however you will loose the most recent ingested data.

Got a checksum exception for /solr/tweets/core_node1/data/tlog/tlog.0000000000000000338 at BP-1464254149-172.17.0.2-1477381671113:blk_1073759482_18761:1024 from xxxxxxxxx:50010\cell \row

.

[#7]

@Jonas Straub HDFS is healthy and solr has been restarted however I'm not sure how to go about doing a manual commit...?

http://localhost:8983/solr/update?commit=true

Please can you tell me how to remove the recent transaction logs? I don't mind losing previously ingested data as this is a test environment. Thanks!

[#8]

@Jonas Straub @Neeraj Sabharwal hi, some guidance on this would be highly appreciated

[#9]

@Raf Mohammed - This is a transaction log file: /solr/tweets/core_node1/data/tlog/tlog.0000000000000000338, so just delete them (I always prefer to move files I'm going to delete to another directory called "TODELETE" in the event that I actually need them and once things look good I then delete them). You may need to restart Solr if you delete these files out from under Solr.

A manual commit should be like this assuming your collection is named "tweets":

http://localhost:8983/solr/tweets/update?commit=true

[#10]

@james.jones Sorry, how can i access and remove these logs please? *newbie* i'm not concerned about losing any previously ingested data, as they were just a collection of tweets from that particular day.

[#11]

@Raf Mohammed - Assuming your index is stored in HDFS rather than the local file system where Solr is running:

hdfs dfs -mkdir /solr/DELETEME_core_node1 
hdfs dfs -mv /solr/tweets/core_node1/data/tlog/tlog.* /solr/DELETEME_core_node1

When you're ready to delete the files, run this command:

hdfs dfs -rm -r /solr/DELETEME_core_node1

[#12]

@james.jones I managed to run those commands without any issues, however when restarting Solr, I'm still getting the same error 😞

29/11/2016, 22:16:42	WARN	null	NativeCodeLoader	Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
29/11/2016, 22:16:44	WARN	null	BlockReaderFactory	I/O error constructing remote block reader.
29/11/2016, 22:16:44	WARN	null	DFSClient	Connection failure: Failed to connect to /172.17.0.2:50010 for file /solr/tweets/core_node1/data/index/segments_9g for block BP-1464254149-172.17.0.2-1477381671113:blk_1073759485_18671:java.io.IOException: Got error for OP_READ_BLOCK, self=/172.17.0.2:33680, remote=/172.17.0.2:50010, for file /solr/tweets/core_node1/data/index/segments_9g, for pool BP-1464254149-172.17.0.2-1477381671113 block 1073759485_18671
29/11/2016, 22:16:44	WARN	null	DFSClient	DFS chooseDataNode: got # 1 IOException, will wait for 673.6703523877359 msec.
29/11/2016, 22:16:45	WARN	null	BlockReaderFactory	I/O error constructing remote block reader.
29/11/2016, 22:16:45	WARN	null	DFSClient	Connection failure: Failed to connect to /172.17.0.2:50010 for file /solr/tweets/core_node1/data/index/segments_9g for block BP-1464254149-172.17.0.2-1477381671113:blk_1073759485_18671:java.io.IOException: Got error for OP_READ_BLOCK, self=/172.17.0.2:33684, remote=/172.17.0.2:50010, for file /solr/tweets/core_node1/data/index/segments_9g, for pool BP-1464254149-172.17.0.2-1477381671113 block 1073759485_18671
29/11/2016, 22:16:45	WARN	null	DFSClient	DFS chooseDataNode: got # 2 IOException, will wait for 6501.220629185886 msec.
29/11/2016, 22:16:52	WARN	null	BlockReaderFactory	I/O error constructing remote block reader.
29/11/2016, 22:16:52	WARN	null	DFSClient	Connection failure: Failed to connect to /172.17.0.2:50010 for file /solr/tweets/core_node1/data/index/segments_9g for block BP-1464254149-172.17.0.2-1477381671113:blk_1073759485_18671:java.io.IOException: Got error for OP_READ_BLOCK, self=/172.17.0.2:33706, remote=/172.17.0.2:50010, for file /solr/tweets/core_node1/data/index/segments_9g, for pool BP-1464254149-172.17.0.2-1477381671113 block 1073759485_18671
29/11/2016, 22:16:52	WARN	null	DFSClient	DFS chooseDataNode: got # 3 IOException, will wait for 6947.415060544425 msec.
29/11/2016, 22:16:59	WARN	null	BlockReaderFactory	I/O error constructing remote block reader.
29/11/2016, 22:16:59	WARN	null	DFSClient	Connection failure: Failed to connect to /172.17.0.2:50010 for file /solr/tweets/core_node1/data/index/segments_9g for block BP-1464254149-172.17.0.2-1477381671113:blk_1073759485_18671:java.io.IOException: Got error for OP_READ_BLOCK, self=/172.17.0.2:33726, remote=/172.17.0.2:50010, for file /solr/tweets/core_node1/data/index/segments_9g, for pool BP-1464254149-172.17.0.2-1477381671113 block 1073759485_18671
29/11/2016, 22:16:59	WARN	null	DFSClient	Could not obtain block: BP-1464254149-172.17.0.2-1477381671113:blk_1073759485_18671 file=/solr/tweets/core_node1/data/index/segments_9g No live nodes contain current block Block locations: 172.17.0.2:50010 Dead nodes: 172.17.0.2:50010. Throwing a BlockMissingException
29/11/2016, 22:17:00	ERROR	null	CoreContainer	Error creating core [tweets_shard1_replica1]: Error opening new searcher
29/11/2016, 22:17:00	ERROR	null	CoreContainer	Error waiting for SolrCore to be created

[#13]

@Raf Mohammed - Also, if you really don't care about the data you can also delete the collection and then recreate it again:

#http://localhost:8983/admin/collections?action=DELETE&name=tweets

[#14]

@james.jones Thanks, I ran that command, but didn't get a confirmation message to say it's deleted, when i re-enter the command it doesn't tell me that it no longer exists. am i missing something?

[#15]

[root@sandbox ~]# http://localhost:8983/solr/tweets/update?commit=true

bash: http://localhost:8983/solr/tweets/update?commit=true: No such file or directory

[#16]

@Raf Mohammed - it looks like something is bad wrong with your HDFS.

So about deleting the collection, I don't recall, but you should be able to look in Solr at the collections and it should be gone. Also if you try to recreate the collection it should tell you that it already exists if it was not deleted. I don't recall if the data will be deleted when the collection is deleted but I think it is.

[#17]

@james.jones - Not sure what else to try, maybe i'm not deleting it from the correct path?

[solr@sandbox root]$ /opt/lucidworks-hdpsearch/solr/bin/solr create -c tweets -d tweet_configs -s 1 -rf 1 -p 8983

Connecting to ZooKeeper at sandbox.hortonworks.com:2181/solr ...

Re-using existing configuration directory tweets

ERROR:

Collection 'tweets' already exists!

Checked collection existence using Collections API command:

http://sandbox.hortonworks.com:8983/solr/admin/collections?action=list

[#18]

screen-shot-2016-11-29-at-224034.png Here's what my Solr UI looks like right now. There must be away to force Solr to overlook this...

[#19 · Accepted solution]

@Raf Mohammed I recreated your error and tested it, although I was not using HDFS, which should not be any different. To delete the corrupt collection you can do these steps.

• shut down solr
• delete all of the shard directories for the collection. (e.g. hdfs dfs -rm -r /solr/tweets/core_node*, or just move them to another directory outside of the "dataDir" directory)
• start solr
• issue the collection DELETE command (e.g. http://localhost:8984/solr/admin/collections?action=DELETE&name=collection2)
• recreate the collection

I recommend shutting down Solr and HDFS gracefully to avoid index corruption. Also, you can create backup copies if you want to recover in the future. Having replicas will also probably help with this situation.

[#20]

@james.jones I got up to step for, but can't proceed due to: [root@sandbox ~]# -bash: http://localhost:8984/solr/admin/collections?action=DELETE: No such file or directory

Would I be ok to recreate the collection?

[#21]

@Raf Mohammed - Click on the "Cloud" icon in the Admin console and your collection should not appear in the graph (or there will be no graph if there are no collections at all). If that's the case, then yes, you can recreate the collection now. It will create the directories it needs.