I am trying to backup the whole cluster consistently. What are different ways to backup and restore Cassandra cluster?
If you are using the DataStax Enterprise version, then the easiest way is to perform the backups and restore using OpsCenter.
If you are using the DataStax Community or open-sourced version of Cassandra, then use nodetool snapshot to create backups of tables and/or keyspaces.
Please bear in mind that SSTables are immutable, i.e. they never change once they are written to disk. So unlike RDBMS data files, SSTables are not updated.
To perform a snapshot cluster-wide, use SSH tools such as pssh to perform parallel snapshots on all nodes.
More information on the snapshot utility is available here.
There are several ways to restore from snapshots. One way is to re-load the data using the sstableloader tool where the data is read back into the cluster. Another way is by copying the SSTable directory from snapshot and running nodetool refresh. Finally, you can replace the existing data with the snapshot and restarting the node.
More information on backups and restores are available here.
Related
Edited after reading nodetool tagged questions.
We take snapshots of our single node cassandra database daily. If I want to restore a snapshot either on that node, or on our staging server which is running a different instance of cassandra, my understanding is I have to:
nodetool disablegossip
nodetool disablebinary
nodetool drain
Copy the sstable files from the snapshot directories to the sstable directories under the keyspace directory.
Run nodetool refresh on each table.
Enable binary & gossip.
Is this sufficient to safely bring the snapshot sstable files in without cassandra overwriting them while I'm doing the refresh?
What is the opposite of nodetool drain?
Another edit: What about sstableloader? Should I use that instead? If so, how? I looked at the "documentation" and am none the wiser.
The steps you outlined isn't quite right. You don't shutdown Cassandra and you shouldn't just copy the files on top of the existing SSTables.
At a high level, the steps to restore table snapshots on a node are:
TRUNCATE the table you want to restore (will remove the SSTables from the data directories).
Copy the SSTables from data/ks_name/table-UUID/snapshots/snapshot_name subdirectory into the "live" data directory data/ks_name/table-UUID.
Run nodetool refresh -- ks_name table_name.
You will need to repeat these steps for each application table you want to restore. NOTE: Do NOT restore system tables, only application tables.
The detailed steps are documented in Restoring from a snapshot in Cassandra.
To restore a snapshot into another cluster, I prefer to refer to this as "cloning". The procedure for cloning snapshots to another cluster depends on whether the source and destination clusters have identical configuration.
If both source and destination clusters are identical, follow the steps I documented here -- https://community.datastax.com/questions/4534/. I've explained what identical configuration means in this post.
If they are not identical, follow the steps I documented here -- https://community.datastax.com/questions/4477/. Cheers!
I am working on a cassandra backup and recovery strategy for our cassandra system and am trying to understand how the backup and sstable recovery works in cassandra. Here are of my observations and related questions (my need is to setup a standby/backup cluster which would become active if the primary cluster goes down.. so I want to keep them in sync in terms of data, so I want to take periodic backups at my active cluster and recover to the standby cluster)
Took a snapshot backup. Dropped a table in cassandra. Stopped cassandra, recovered from the snapshot backup (copied the sstables to the data/ folder), started cassandra. Ran cqlsh on the node, and I still do not see the table created. Should this work? Am I missing any step ?
In the above scenario, I then tried to re-setup the schema (I take backup of the schema in the snapshot) using the cql commant source . This created the table for me. However it creates a "new version" of table for me. When I recover the snapshot has the older version (different uuid labelled folders for table). After recovery, I still see no data in the table. Possibly because I created a new table?
I was finally able to recover data after running nodetool repair and using sstableloader to restore table data from another node in the cluster.
My question is
a. what is the right way to setup a new (blank- no schema) cluster from a snapshot? How do you setup the schema and recover data?
b. how do you restore a cluster from a backup with table alterations. How do you bring a cluster running an older version of schema to a newer version of schema when recovering from a backup (snapshot or incremental)?
(NOTE: cassandra newbie here)
So if you want to restore a snapshot, you need to copy the snapshot files back to the sstable directory and then run: nodetool refresh. You can read:
https://docs.datastax.com/en/dse/5.1/dse-admin/datastax_enterprise/operations/opsBackupSnapshotRestore.html
for more information. Also, there are 3rd party tools that can back up your data and then restore it as it was at the time of the backup. We use a tool: Cohesity (formally Talena/Imanis). It has a lot of capabilities (refreshing A to B, restore/rename, etc.). There are other popular ones as well. All of them have costs associated with them.
Hopefully that helps?
-Jim
Offsite backups for Cassandra seem like a challenging thing. You basically have to make yet another copy of ALL your data, including the copies of data that exist due to the replication factor. Snapshots make backups easy when you don't mind storing it on the same disk that your node already uses. I'm curious - in the event of a catastrophic failure of this disk, is it possible to recover the node using the nodes that the data was replicated to?
Yes, you can restore data on crashed node using a procedure in documentation - Replacing a dead node or dead seed node. It's for Cassandra 3.x, please pick your Cassandra version from a drop-down menu on the top of the page.
But please note that you still need to do backups if your data is valuable. If you using AWS you can use this project to backup Cassandra to S3 storage.
If you are looking for offsite or off-host backups, you can also look at opscenter from Datastax or Talena software (my company). Both provide you the ability to backup your database locally or to S3. As you may expect, you also have the ability to restore data in case of hardware failures, user errors or logical corruptions which the replicas will not protect you against.
Yes, it is possible. Just execute in terminal "nodetool repair" on the node with missed data. It can take a lot of time. Also I would recommend execute repair operation on each node every month to keep your data always replicated because cassandra does not repairs data automatically (for example after node(s) falling).
We have set up a backup/restore procedure for our Cassandra production environment via snapshots. The snapshot files, schema and token ring information are copied to S3.
The production cluster is a 3-node-cluster with a replication factor of 3.
For development and test, I would like to restore the snapshots from production into separated clusters. To save money and to keep maintenance easy, it would be nice to restore only the snapshot from one production node. Since we are using a replication factor of 3 in a 3-node-cluster, each snapshot should have all rows. Consistency is also not important for our use-case.
Is it possible (and how) to restore only a single snapshot?
All of your data should exist on all 3 nodes so copying the sstables from any 1 node to your test cluster should be sufficient. Making sure theres a recent repair beforehand may be good idea if worried about consistency.
First create the same schema on the test cluster. Then you can simply take a snapshot with nodetool snapshot -t cloneme. Once complete, copy all the sstables from the folder that is created (cloneme) into the equivalent tables folder on your test cluster. Then run nodetool refresh.
It gets much more complicated if you have a different topology (more nodes, different RF) but since your going with "every node has all the data" its pretty trivial.
Worth mentioning that OpsCenter has a feature to automate the copying of a backup to other clusters.
We have a 10 node Cassandra cluster. We configured a repair in Opscenter. We find there is a backups folder created for every table in Opscenter keyspace. It keeps growing huge. Is there a solution to this, or do we manually delete the data in each backups folder?
First off, Backups are different from snapshots - you can take a look at the backup documentation for OpsCenter to learn more.
Incremental backups:
From the datastax docs -
When incremental backups are enabled (disabled by default), Cassandra
hard-links each flushed SSTable to a backups directory under the
keyspace data directory. This allows storing backups offsite without
transferring entire snapshots. Also, incremental backups combine with
snapshots to provide a dependable, up-to-date backup mechanism.
...
As with snapshots, Cassandra does not automatically clear
incremental backup files. DataStax recommends setting up a process to
clear incremental backup hard-links each time a new snapshot is
created.
You must have turned on incremental backups by setting incremental_backups to true in cassandra yaml.
If you are interested in a backup strategy, I recommend you use the OpsCenter Backup Service instead. That way, you're able to control granularly which keyspace you want to back up and push your files to S3.
Snapshots
Snapshots are hardlinks to old (no longer used) SSTables. Snapshots protect you from yourself. For example you accidentally truncate the wrong keyspace, you'll still have a snapshot for that table that you can bring back. There are some cases when you have too many snapshots, there's a couple of things you can do:
Don't run Sync repairs
This is related to repairs because synchronous repairs generate a Snapshot each time they run. In order to avoid this, you should run parallel repairs instead (-par flag or by setting the number of repairs in the opscenter config file note below)
Clear your snapshots
If you have too many snapshots and need to free up space (maybe once you have backed them up to S3 or glacier or something) go ahead and use nodetool clearsnapshots to delete them. This will free up space. You can also go in and remove them manually from your file system but nodetool clearsnapshots removes the risk of rm -rf ing the wrong thing.
Note: You may also be running repairs too fast if you don't have a ton of data (check my response to this other SO question for an explanation and the repair service config levers).