Cleaning Search Node Disk

Full Text Search
1

I’m running out of space on a search node.

In /opt/couchbase/var/lib/couchbase/data/@fts I see that there are many indexes of the following format:

indexName_alphanumericString1_alphanumericString2

On each line, the indexName and alphanumericString1 are the same, but alphanumericString2 is different.

Are these snapshots of the same index? If so, is there a way to dump old snapshots?

Thanks.

Full Example with Dummy Data:

    4.0 GiB [##########] /searchable-index_5bb73sdjkjfr23ao_13ca523f.pindex
    3.9 GiB [######### ] /searchable-index_5bb73sdjkjfr23ao_f4e0d46m.pindex
    3.8 GiB [######### ] /searchable-index_5bb73sdjkjfr23ao_548202l2.pindex
    3.6 GiB [########  ] /searchable-index_5bb73sdjkjfr23ao_18qazd87.pindex
    3.4 GiB [########  ] /searchable-index_5bb73sdjkjfr23ao_aa574cv8.pindex
    3.4 GiB [########  ] /searchable-index_5bb73sdjkjfr23ao_m8dbdsdf.pindex
    1.7 MiB [          ] /dumps
    4.0 KiB [          ]  cbft.uuid
2

Hi @joes ,

-Each of these is the storage for the index partitions. By default, there will be 6 index partitions for an index and you may explore and edit that in the index definition.
-Editing the partition count result in a rebuild of the index.
-You may explore ways to tune your index definition to minimize disk usage.
Full-Text Search Indexing Best Practices & Tips - Part 1 and 7 Useful Tips for Your Full-Text Search Production Systems | The Couchbase Blog

Cheers!

1 Like