On Tue, Sep 20, 2011 at 7:01 AM, Ryan Zezeski <[email protected]> wrote:
> Index entries are removed by the Search hook during object deletion. It
> sounds like you removed the Search hook, then removed the objects, and now
> you have entires in your Search index for non-existing objects. If you
> don't care about losing **all** of your indexes then you should be able to
> take the cluster down, delete all the `data/merge_index` dirs, and then
> restart the cluster. At that point all your objects should be in tact but
> the indexes should be gone. It's important that you take the cluster down
> for this to work as merge_index keeps some data in memory that can survive.
>
>
Ryan,
Thanks. Yes, this is what I ended up doing. That said, I think being able
to drop an index without deleting all other indexes or deleting your objects
from the KV store, should be a basic operations feature. I created an
enhancement request for it.
> If you know the object ids then you could delete only the necessary indexes
> via a small bit of Erlang at the riak console via `search:delete_docs/1`.
>
> search:delete_docs([{<<"bucket">>, <<"key1">>}, ...])
>
That may come in handy. Thanks.
Elias Levy
_______________________________________________
riak-users mailing list
[email protected]
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
