Hi Luciano,

It is not a must, as, after the node is upgraded all the new sstables will be 
created as oa sstables. Then old nb sstables will get compacted away 
eventually, but this eventually could be a long time in the future.

So I would recommend running the sstable upgrade, even if you wait to schedule 
it during a time of lower load on the system.

Regards 

Paul 

Sent from my iPhone

> On 18 Mar 2025, at 18:47, Luciano Greiner <luciano.grei...@gmail.com> wrote:
> 
> One thing I am not really sure is if upgrading sstables is really a
> must after the Cassandra 5 upgrade.
> 
> Thanks
> 
> Luciano Greiner
> 
> 
>> On Sun, Mar 16, 2025 at 2:33 PM Paul Chandler <p...@redshots.com> wrote:
>> 
>> Yes, that should sort it out.
>> 
>> Regards
>> 
>> Paul
>> Sent from my iPhone
>> 
>>>> On 16 Mar 2025, at 18:34, Luciano Greiner <luciano.grei...@gmail.com> 
>>>> wrote:
>>> 
>>> Thank you Paul!
>>> 
>>> So should I restart the nodes with UPGRADING mode and run the
>>> upgradesstables again?
>>> 
>>> Thank you!
>>> 
>>> Luciano Greiner
>>> (54) 996309845
>>> 
>>>> On Sun, Mar 16, 2025 at 2:58 AM Paul Chandler <p...@redshots.com> wrote:
>>>> 
>>>> Hi Luciano,
>>>> 
>>>> It sounds like you could have the storage_compatibility_mode set to the 
>>>> default CASSANDRA_4 value, check this and change it to UPGRADING or NONE.
>>>> 
>>>> Full details can be found in the Cassandra.yaml 
>>>> https://cassandra.apache.org/doc/latest/cassandra/managing/configuration/cass_yaml_file.html#storage_compatibility_mode
>>>> 
>>>> Regards
>>>> 
>>>> Paul
>>>> 
>>>> Sent from my iPhone
>>>> 
>>>> On 16 Mar 2025, at 07:48, Luciano Greiner <luciano.grei...@gmail.com> 
>>>> wrote:
>>>> 
>>>> Hi,
>>>> 
>>>> We recently upgraded our clusters from 4.1.5 to 5.0.3 and now I’m
>>>> trying to migrate SSTable files to the new oa-* format, but it’s not
>>>> working as I expected.
>>>> 
>>>> What I Tried:
>>>> 
>>>> nodetool flush + upgradesstables → Completed quickly with success
>>>> messages, but no SSTables were rewritten.
>>>> nodetool upgradesstables -a → Triggered compactions, but new SSTables
>>>> are still in the old nb-* format.
>>>> 
>>>> What’s the recommended approach to ensure SSTables migrate to oa-*? Is
>>>> it really necessary?
>>>> 
>>>> Thanks!
>>>> 
>>>> Luciano
>> 

Reply via email to