Alternative prod backup strategies?


#1

Hi,

We’ve got a production couchbase cluster (comm ed 2.2.0) with 5 nodes, around 150GB data & 6M records. We don’t yet have a backup solution. We’ve tried cbbackup and XDCR. In both cases, the prod cluster began to fail (clients began to time out requesting data). I suspect the nodes were cpu constrained (due to other processes on those machines). In both cases, the backup was to a separate machine outside of the prod cluster.

Either way, we still need to back up our data. We’re new to couchbase, so maybe we’ve done something wrong with the existing cbbackup or XDCR options.

Also, we’re running on AWS, couchbase manual installed on m3.large instances using EBS (ssd, gp2) for data.

Any recommendations or direction would be much appreciated!

Thanks,
Travis


Java SDK 1.4 failing during cbbackup
#2

That’s not expected. I believe we recommend xlarge at the minimum, check the docs. Note that there was some discussion about node types at the Couchbase New York conference in June you may be able to find.

I think my immediate recommendation would be to snapshot the EBS volumes so you’ll at least have the data files. XDCR can use some resources, but I’d be surprised if cbbackup would cause this on a supported instance type.