cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Léo FERLIN SUTTON <lfer...@mailjet.com.INVALID>
Subject Speed up compaction
Date Thu, 13 Jun 2019 09:27:40 GMT
I am currently noticing very very slow compactions on my cluster and
wondering if there is any way to speed things up.

Right now I have this compaction currently running :

   60c1cfc0-8da7-11e9-bc08-3546c703a280        Compaction   keyspace1
> table1     8.77 GB   1.71 TB   bytes      0.50%


The speed is abysmal, it only progresses by "0.24%" every _hour_. At this
rate it is going to take weeks...

When I look at the cpu load and disk load I see :
Never more than 10MB/s read and write combined.
Never more than 6 load average, most of the cpu time is spent idle.

## Hardware configuration :

The node is a n1-highmem-8 (8 vCPUs, 52 GB memory) instance on google cloud
engine.

The disk performances should be around :
15,000.00 random IOPS (read write)
800MB/s Read
400 MB/s Write

## Cassandra configuration :
4 concurrent_compactors
Current compaction throughput: 150 MB/s
Concurrent reads/write are both set to 128.

I have also temporarily stopped every repair operations.

Any ideas about how I can speed this up ?

Regards,

Leo

Mime
View raw message