Databases
StackOverflow2013
Postsdbo
POUneven data size on some cassandra nodes after extending and cleanup
 

Note that there are some explanatory texts on larger screens.

plurals
  1. POUneven data size on some cassandra nodes after extending and cleanup
    primarykey
    data
    text
    <p>1) We had cluster of 10 nodes, recently we added 20 more nodes to the cluster.</p> <p>2) After addition we ran cleanup on all the necessary nodes.</p> <p>3) In ring status "Effective-Owership" is properly balanced but "load" on two machine is different than rest of the machines.</p> <pre><code>rack1 Up Normal 196.38 GB 6.67% rack2 Up Normal 195.33 GB 6.67% rack1 Up Normal 191.57 GB 6.67% rack2 Up Normal 197.83 GB 6.67% rack1 Up Normal 190.92 GB 6.67% rack2 Up Normal 194.59 GB 6.67% rack1 Up Normal 195.66 GB 6.67% rack2 Up Normal 191.45 GB 6.67% rack1 Up Normal 197.13 GB 6.67% rack2 Up Normal 196.19 GB 6.67% rack1 Up Normal 195.39 GB 6.67% rack2 Up Normal 199.35 GB 6.67% rack1 Up Normal 197.71 GB 6.67% rack2 Up Normal 194.22 GB 6.67% rack1 Up Normal 192.83 GB 6.67% rack2 Up Normal 197.17 GB 6.67% rack1 Up Normal 192.61 GB 6.67% rack2 Up Normal 193.88 GB 6.67% rack1 Up Normal 197.3 GB 6.67% rack2 Up Normal 196.74 GB 6.67% rack1 Up Normal 194.89 GB 6.67% rack2 Up Normal 198.47 GB 6.67% rack1 Up Normal 197.26 GB 6.67% rack2 Up Normal 345.34 GB 6.67% rack1 Up Normal 195.68 GB 6.67% rack2 Up Normal 263.23 GB 6.67% rack1 Up Normal 190.72 GB 6.67% rack2 Up Normal 198.98 GB 6.67% rack1 Up Normal 194.22 GB 6.67% rack2 Up Normal 191.95 GB 6.67% </code></pre> <p>4) On one machine load is 345GB and on other machine it is 263GB while on rest of the machine it is around 195GB.</p> <p>5) We are using Cassandra-1.1.0 and I have run cleanup on these machine twice but it is not helping.</p> <p>Any Idea how could I balance this cluster with same load on each node?</p>
    singulars
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PTQuestion
    1. USsamarth
    1. USsamarth
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. COIs your Cassandra cluster configured to use ByteOrderedPartitioner or RandomPartitioner? You can find out by running cassandra-cli and then running the "describe cluster" command. ByteOrderedPartitioner can have the side effect of uneven distribution of data across nodes in a cluster, depending on usage patterns.
      singulars
      1. POUneven data size on some cassandra nodes after extending and cleanup
      1. USChris Nauroth
    2. COwe are using RandomPartitioner only and cluster data load was perfectly balanced before.Even now "Effective-Owership" is balanced but data load is uneven.
      singulars
      1. POUneven data size on some cassandra nodes after extending and cleanup
      1. USsamarth
    3. COIs there any chance that something about your data set has changed such that keys do not naturally fall into an even distribution? To clarify, consider the example of using letters of the alphabet as keys, words starting with those letters as column names, and counts of occurrences of those words in a document as column values. This arrangement likely would yield an uneven distribution of data across nodes, because there are far more words that start with 'T' than words that start with 'Z'.
      singulars
      1. POUneven data size on some cassandra nodes after extending and cleanup
      1. USChris Nauroth
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload