0

Someone accidentally deleted several records from Cassandra's table in the cluster. I want to restore the last snapshot to another copy table with the same structure within the cluster.

What is the best way to do that?

Cassandra v4.0.4

Super Kai - Kazuya Ito
  • 213
  • 1
  • 8
EdiM
  • 51
  • 6

3 Answers3

0

You could use SSTABLELOADER or NODETOOL IMPORT. Documentation for those sources are below.

https://cassandra.apache.org/doc/4.1/cassandra/operating/bulk_loading.html#using-sstableloader

https://cassandra.apache.org/doc/4.1/cassandra/operating/bulk_loading.html#using-nodetool-import

Matthew
  • 11
0

I've previously documented steps on how to clone snapshots to another cluster in How do I restore Cassandra snapshots to another cluster with identical configuration?. The procedure for cloning a snapshot to a new table is pretty much identical using the "refresh" method.

The high level steps are:

STEP 1 - Create the table with a new name using the schema file in the snapshot.

STEP 2 - On the first node, copy the contents of the snapshot directory to the corresponding data directory of the destination table.

STEP 3 - Force Cassandra to load the copied SSTables from disk with:

$ nodetool import -- ks_name new_table_name

STEP 4 - Check the Cassandra system.log to verify that the new SSTables were opened.

STEP 5 - Repeat steps 2 to 4 on the next node until all nodes are done.

See the background information and additional details in the post I linked above. Cheers!

Erick Ramirez
  • 4,590
  • 1
  • 8
  • 30
0

Thank you for your reply, @Erick .
I experimented with your recommendation, putting it to the test using two tables I constructed with identical structures edi_test_keyspace.t1 and edi_test_keyspace.t1_copy .
I inserted a few records to t1, and then I truncated the t1 table (in order to create an automated snapshot).
I copied the files from the truncated snapshot t1 directory to the corresponding data directory of t1_copy table
Then I ran the command nodetool import edi_test_keyspace t1_copy but I got the following error nodetool: import requires keyspace, table name and directories See 'nodetool help' or 'nodetool help <command>.'
the command works only after I specify the t1_copy directory path nodetool import edi_test_keyspace t1_copy /data/cassandra/edi_test_keyspace/t1_copy-d0659b50ffd911ee9b3c6bc1eefe57b6

EdiM
  • 51
  • 6