"Failed to compact" in RiakSearch

Morten Siebuhr sbhr+lists at sbhr.dk
Mon Apr 18 07:01:39 EDT 2011


On Fri, Apr 15, 2011 at 4:40 PM, Rusty Klophaus <rusty at basho.com> wrote:
> Hi Morten,
> It looks like at least one of the offending fields is "photos_value", which,
> as you said, should be skipped according to your schema. This makes me think
> that either the schema isn't set correctly, or that one or more nodes is
> caching an old value of the schema.
> Can you try running "search-cmd show-schema INDEXNAME" on each node to
> verify that the schema is set correctly? Also, you can run "search-cmd
> clear-schema-cache" to clear the schema cache across all nodes.

Bingo!

It turned out *all* of the nodes were using the default schema...

(I the schema on the "nosql1" in the beginning, believing it would be
replicated to the other nodes.)

> Also, the number of segments is *way* higher than it should be, and this is
> the reason for the "too many DB tables" error. It appears that these
> problems are linked, the compaction errors are preventing the system from
> combining segments, leading to a large number of segments, so solving one
> problem should solve the other.

We'll try to re-add our data and see what happens...

/Siebuhr




More information about the riak-users mailing list