I am looking for confirmation that my Cassandra backup and restore procedures are sound and I am not missing anything. Can you please confirm, or tell me if something is incorrect/missing?
Backups:
I run daily full backups of the keyspaces I care about, via "nodetool snapshot keyspace_name -t current_timestamp". After the snapshot has been taken, I copy the data to a mounted disk, dedicated to backups, then do a "nodetool clearsnapshot $keyspace_name -t $current_timestamp"
I also run hourly incremental backups - executing a "nodetool flush keyspace_name" and then moving files from the backup directory of each keyspace, into the backup mountpoint
Restore:
So far, the only valid way I have found to do a restore (and tested/confirmed) is to do this, on ALL Cassandra nodes in the cluster:
Stop Cassandra
Clear the commitlog *.log files
Clear the *.db files from the table I want to restore
Copy the snapshot/full backup files into that directory
Copy any incremental files I need to (I have not tested with multiple incrementals, but I am assuming I will have to overlay the files, in sequence from oldest to newest)
Start Cassandra
On one of the nodes, run a "nodetool repair keyspace_name"
So my questions are:
Does the above backup and restore strategy seem valid? Are any steps inaccurate or anything missing?
Is there a way to do this without stopping Cassandra on EVERY node? For example, is there a way to restore the data on ONE node, then somehow make it "authoritative"? I tried this, and, as expected, since the restored data is older, the data on the other nodes (which is newer) overwrites in when they sync up during repair.
Thank you!
There's two ways to restore Cassandra backups without restarting C*:
Copy the files into place, then run "nodetool refresh". This has the caveat that the rows will still be older than tombstones. So if you're trying to restore deleted data, it won't do what you want. It also only applies to the local server (you'll want to repair after)
Use "sstableloader". This will load data to all nodes. You'll need to make sure you have the sstables from a complete replica, which may mean loading the sstables from multiple nodes. Added bonus, this works even if the cluster size has changed. I'm not sure if ordering matters here (that is, I don't know if row timestamps are preserved through the load or if they're redefined during load)
Related
I searched through the internet a lot and saw a lot of ways to backup and restore a Cassandra cluster, such as nodetool snapshot and Medusa. but my question is that can I use dsbulk to backup a Cassandra cluster. What are its limitations? Why doesn't anyone suggest that?
It's possible to use it in some cases, but it's not practical because (that are primary, list could be bigger):
DSBulk put an additional load onto the cluster nodes because it's going through the standard read path. In contrast to that nodetool snapshot just create a hardlinks to the files with data, no additional load to the nodes
It's harder to implement incremental backups with DSBulk - you need to come with condition for SELECT that will find only data that changed since the last backup, so you need to have timestamp column, because you can't do the WHERE condition on the value of writetime function. Plus it will require rescanning of whole data anyway. Plus it's impossible to find what data were deleted. With nodetool snapshot, you just compare what files has changed since last backup, and backup only them.
Currently we have two options to take data back up of the tables in a Cassandra keyspace. We can either user nodetool commands or use the copy command from the cqlsh terminal.
1) What are the differences between these commands ?
2) Which one is most appropriate ?
3) Also if we are using nodetool to take backup we would generally flush the data from mem tables to sstables before we issue the nodetool snapshot command. So my question is should we employ the same techinque of flushing the data if we use the cqlsh copy command ?
Any help is appreciated.
Thanks very much.
GREAT question!
1) What are the differences between these commands ?
Running a nodetool snapshot creates a hard-link to the SSTable files on the requested keyspace. It's the same as running this from the (Linux) command line:
ln {source} {link}
A cqlsh COPY is essentially the same as doing a SELECT * FROM on a table. It'll create a text file with the table's data in whichever format you have specified.
In terms of their difference from a backup context, a file created using cqlsh COPY will contain data from all nodes. Whereas nodetool snapshot needs to be run on each node in the cluster. In clusters where the number of nodes is greater than the replication factor, each snapshot will only be valid for the node which it was taken on.
2) Which one is most appropriate ?
It depends on what you're trying to do. If you simply need backups for a node/cluster, then nodetool snapshot is the way to go. If you're trying to export/import data into a new table or cluster, then COPY is the better approach.
Also worth noting, cqlsh COPY takes a while to run (depending on the amount of data in a table), and can be subject to timeouts if not properly configured. nodetool snapshot is nigh instantaneous; although the process of compressing and SCPing snapshot files to an off-cluster instance will take some time.
3) Should we employ the same technique of flushing the data if we use the cqlsh copy command ?
No, that's not necessary. As cqlsh COPY works just like a SELECT, it will follow the normal Cassandra read path, which will check structures both in RAM and on-disk.
nodetool snapshot is good approach for any amount of data and it creates a hard link within seconds.copy command will take much time because and depends upon the size of data and cluster. for less data and testing you may use copy command but for production nodetool snapshot is recommended.
Currently, I am bench marking Cassandra database using YCSB framework. During this time I have performed (batch) insertion and deletion of the data quite regularly.
I am using Truncate command to delete keyspace rows. However, I am noticing that my Cassandra data directory swells up as the experiments.
I have checked and can confirm that even there is no data in the keystore when I checked the size of data directory. Is there a way to initialize a process so that Cassandra automatically release the stored space, or does it happen over time.
When you use Truncate cassandra will create snapshots of your data.
To disable it you will have to set auto_snapshot: false in cassandra.yaml file.
If you are using Delete, then cassandra use tombstone,i.e your data will not get deleted immediately. Data will get deleted once compaction is ran.
To remove previous snapshots one can use nodetool snapshot command.
I am also looking for incremental/point-in-time backup / restore solution.
I have three Cassandra nodes,I enabled incremental backup , and i tried copy one day's SSTable files from backups folder on one node to a new cassandra cluster /data folder, then it works, but i have three node, and the name on all three node are same, i dont' know how to restore the incremental backup files from all the three nodes.
You comments are really appreciated !
One simple solution is to config new cluster with exact same node, and everything can work.
But if i want to replay the data with less nodes, like only one, so i have to take a complete snapshot ?
I have enabled the incremental backup in the cassandra.yaml file. As I know when we enable incremental backups, cassandra will backup the data (in backups directory) only when the memtable is flushed. But what if the memtable is yet to be flushed? I won't be able to get the incremental backup right?. I know that for the memtable to be flushed there are certain conditions to be met such as time interval or memtable space. My question is how do I modify this so that even if I enter one record after the last snapshot, I can still backup entire data along with that latest entry?
Consider this example
Take the snapshot.
Clear incremental backup (backups directory)
Enter a record to a table.
Check for the incremental backup in backups directory. It is still empty.
Now how do I backup the record which is written after the last snapshot?In general how do we backup the entire upto-date data unless we take the snapshot?
You can flush the files manually with nodetool flush just before taking the backup. That way you'll always have the latest memtable flushed.
nodetool docs
If you want to backup a cluster without taking a snapshot you can do it by simply saving everything under /data folder from every node (this includes mainly the .db files stats files etc).
In order to not override files you should store it with the token information as well.
When you want to restore from this backup, you should spin up a cluster with the same number of nodes, and simply copy the data, one-to-one from each backed-up node to a restored node. Pay attention that you'll have to modify cassandra.yaml to include the relevant token in cassandra.yaml (as well as the peers/seeds/etc) for each restored node.
After all the data is copied, you can start C* process on all the nodes.