Skip to main content



HBase Utility - Merging Regions in HBase

 

Growing HBase Cluster, and difficulty to get physical hardware is something that every enterprise deals with...

And sometimes, the amount of ingest starts putting pressure on components before new hosts can be added. As, I write this post our cluster was running with 46 nodes and each node having 600 regions per server. This is bad for performance.

You can use the following formula to estimate the number of regions for a RegionServer:

(regionserver_memory_size) * (memstore_fraction) / ((memstore_size) * (num_column_families))

Reference https://docs.cloudera.com/HDPDocuments/HDP3/HDP-3.0.0/hbase-data-access/content/memstore-size-regionservers.html


We noticed that HBase doesn't have automatic process to reduce and merge regions. As over the time, many small size or empty regions are formed on cluster which degrades the performance. 

While researching to cope up with this problem: we came across following scripts - 

  • https://appsintheopen.com/posts/51-merge-empty-hbase-regions
  • https://blg.robot-house.us/posts/merging-regions-in-hbase/
We analyzed that - 
  • It is merging empty regions only.
  • Later ruby script creates occasional overlaps. 
  • It doesn't have check for Degenerated, Splitting, Meta Region.
  • It doesn't have size based check, such that merge should not lead to HStore size greater then hregion.max.filesize.

  • It is using deprecated Java functions.
  • It is merging same region again which many a times lead to error where region is not found.

So, we came up with enhanced Java Utility, with all the checks in place. Such that it helps merging adjacent regions - 

Note - HBase region adjacent means lexicographically adjacent. That is end key of one region is start key of another, which can be hosted on different region servers.


Command to run Merge Script - 
 

sh merge_hbase_table.sh "<HBase Table Name>" true

Java Source code location - 

https://github.com/dinesh028/engineering/blob/master/HBaseUtility/src/main/java/com/aquaifer/HBaseMergePartitions.java 


This Utility is utilizes Online Merge https://docs.cloudera.com/runtime/7.2.10/managing-hbase/topics/hbase-online-merge.html , which issues Asynchronous command to Merge Adjacent Regions.


This Utility helped us bring down 18652 regions to 14500 in a few minutes. But, use this utility in off-hours as Merge Regions may invoke minor and major compactions, which in-turn can lead to performance degradation.


Another way is to set appropriate SPLIT_PLOICY on HBaseTable - 

alter '<TABLE_NAME>', {'SPLIT_POLICY' => '<SPLIT_POLICY>'}


<SPLIT_POLICY> can be replaced with one of below - 

  • org.apache.hadoop.hbase.regionserver.ConstantSizeRegionSplitPolicy

A RegionSplitPolicy implementation which splits a region as soon as any of its store files exceeds a maximum configurable size.

  • org.apache.hadoop.hbase.regionserver.IncreasingToUpperBoundRegionSplitPolicy

Split size is the number of regions that are on this server that all are of the same table, cubed, times 2x the region flush size OR the maximum region split size, whichever is smaller.

For example, if the flush size is 128MB, then after two flushes (256MB) we will split which will make two regions that will split when their size is 2^3 * 128MB*2 = 2048MB.

If one of these regions splits, then there are three regions and now the split size is 3^3 * 128MB*2 = 6912MB, and so on until we reach the configured maximum file size and then from there on out, we'll use that.

  • org.apache.hadoop.hbase.regionserver.BusyRegionSplitPolicy

This class represents a split policy which makes the split decision based on how busy a region is. The metric that is used here is the fraction of total write requests that are blocked due to high memstore utilization. This fractional rate is calculated over a running window of "hbase.busy.policy.aggWindow" milliseconds. The rate is a time-weighted aggregated average of the rate in the current window and the true average rate in the previous window.

  • org.apache.hadoop.hbase.regionserver.DelimitedKeyPrefixRegionSplitPolicy
  • org.apache.hadoop.hbase.regionserver.KeyPrefixRegionSplitPolicy
  • org.apache.hadoop.hbase.regionserver.SteppingSplitPolicy


Comments

Post a Comment

Popular posts

Python [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: Missing Authority Key Identifier

  Error requests.exceptions.SSLError: HTTPSConnectionPool  Max retries exceeded with url:  (Caused by SSLError(SSLCertVerificationError(1, '[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: Missing Authority Key Identifier (_ssl.c:1028)'))). Analysis & Solution Recently, we updated from Python 3.11 to 3.13, which resulted in error above. We did verify AKI = SKI in chain of certificates. Also, imported chain of certificates into certifi. Nothing worked for us. Seemingly, it is a bug with Python 3.13. So, we downgraded to Python 3.12 and it started working. Other problems and solution -  '[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: self-signed certificate in certificate chain (_ssl.c:1006)'))) solution  pip install pip-system-certs [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: certificate has expired  (_ssl.c:1006) solution  1# openssl s_client -showcerts -connect  signin.aws.amazon.com:443  </dev/...
Post a Comment
Read more




Spring MongoDB Rest API not returning response in 90 seconds which is leading to client timeout

  We have Spring Boot  Rest API deployed in Kubernetes cluster which integrates with MongoDB to fetch the data.  MongoDB is fed with data by a real time Spark & NiFi job.  Our clients complained that for a request what they send they don't have response within 90 seconds. Consider it like an OMS ( Order ManagEment System).  On further analysis, we found that Spark & NiFi processing is happenning within 10 seconds after consuming response data from Kafka. Thus, initally out thought was that it due to delay from upstream to produce data in to Kafka.  Thankfully, our data had create / request  timestamp, and when response was received, and when response was inserted into MongoDB. Subtracting response insert time from request time seemed to be well within 90 seconds. But, still client did timeout on not seeing a response within 90 seconds. This led to confusion on our side.  But, then we realized it was due to Read Preference . We updated this...
Post a Comment
Read more




MongoDB Regex Query taking more time in Production but same query perform well in UAT

   We came across a situation where-in, MongoDB Query was taking more time in Production like 10 seconds and 4.2 seconds but same query performed well in UAT taking under 400 ms. The very first thought that was evident to us that it is because of amount of data which differed in UAT and Production. Then we ran following to see the execution plan -   db.collection.aggregate(<queries>).explain() This gave us Winning and Rejected Plans. Under which, we analyzed that although it was using 'IXSCAN.' But, it was incorrect index- as we had one compound index built on time field and other fields, and there was other index just on time field for TTL purposes. Winning plan picked TTL index rather than compound index. Thus, we dropped TTL index and built TTL index on a different time field.  That got our query performance time from 10 seconds to 726 ms. Also, for other query the performance came down from 8 seconds to 4.3 seconds. Then, we ran following -  ...
Post a Comment
Read more




What is Leadership

 
Post a Comment
Read more




Spark MongoDB Connector Not leading to correct count or data while reading

  We are using Scala 2.11 , Spark 2.4 and Spark MongoDB Connector 2.4.4 Use Case 1 - We wanted to read a Shareded Mongo Collection and copy its data to another Mongo Collection. We noticed that after Spark Job successful completion. Output MongoDB did not had many records. Use Case 2 -  We read a MongoDB collection and doing count on dataframe lead to different count on each execution. Analysis,  We realized that MongoDB Spark Connector is missing data on bulk read as a dataframe. We tried various partitioner, listed on page -  https://www.mongodb.com/docs/spark-connector/v2.4/configuration/  But, none of them worked for us. Finally, we tried  MongoShardedPartitioner  this lead to constant count on each execution. But, it was greater than the actual count of records on the collection. This seems to be limitation with MongoDB Spark Connector. But,  MongoShardedPartitioner  seemed closest possible solution to this kind of situation. But, it per...
Post a Comment
Read more