You should not bootstrap with mixed cluster. First you should upgrade and
then bootstrap a new node. If you are not able to upgrade due to disk space
constraints while running upgradesstable, then
1. Reduce the number of threads for upgradesstable (--jobs parameter). By
default it is 2. You can do it one.
2. Run upgradesstable with keyspace/table
On Sat, May 9, 2020 at 9:51 PM Surbhi Gupta
wrote:
> Hi,
>
> We are facing some issue in bootstrapping new node in 3.11.0 and
> bootstrapping is failing.
> We have two tasks here :
> 1. Expand the cluster (Due to disk concern and dropped mutation)
> 2. Upgrade the cluster from 3.11.0 to 3.11.5 because of various bugs we
> are hitting in 3.11.0 .
>
> So my question here is :
> 1. Can we add new node with 3.11.5 rpm on a 3.11.0 cluster ?
> 2. Because there is a sstable format change from mc to md , is there any
> issue in bootstrapping a node with 3.11.5 version ?
>
> We are not able to upgrade the cluster first and add nodes because of the
> disk concerns . upgradesstable will need space because of sstable format
> change.
>
> Please advise.
>
> Thanks
> Surbhi
>
>
