Bulk documents update using N1QL update

Couchbase Server
1

Hi,

I tried updating bulk documents using the n1ql update query. The issue I have faced was that only a subset of documents was getting updated in the first try, a little more in the second try and so on without change the condition in WHERE part. Is this a bug in couchbase server or a limit on the number of documents getting updated in one single query. I am using couchbase version
Enterprise Edition 5.1.0 build 5552.

Thanks
Shubham Goel

1 Like
2

You need to be more specific. Did you got error. what is error. Post actual update with explain and index definition.

3

I did not get any error, I got usual response i.e. empty array after executing the update query.

QUERY: Update bucket_name SET channel=[‘value’]
where companyId = ‘value2’
AND locationId = ‘value3’" AND Meta(bucket_name).id NOT LIKE “%_sync%”

Index Definition: I did not specifically had an index for this query, because I was not concerned with latency, I wanted to execute a backward migration script in background.

4 
CREATE INDEX ix1 ON  bucket_name(companyId, locationId) 
WHERE META().id NOT LIKE "%_sync%";

Also check Status of the update

5

Can you tell why partial mutations without creating index, from my understanding indexing is used to avoid latency? And one more thing I tried now without indexing on query workbench it worked fine, sometimes it causes partial mutations, and sometimes it works as excpected.

6

Without index it might use primary index. get all documents and apply filters and discard entries.
Lot of unnecessary work. Also check any timeout . Any index is eventual consistent if index is not up to date on mutations those might not updated.

7
  1. Got the primary index point.
  2. index is not up to date - how to ensure the index is up to date, do primary index not get update ?
  3. Also check any timeout - what is the problem in timeout, and I do have any timeout, you are basically referring timeout as “execute the query after fixed millisecond”?
8

Couchbase index are eventual consistence. If you want you use scan_consistency (check documentation).
While executing query user can set timeout. status of query gives if any tieout.

9

I think I will have to read the documentation as there is a gap in my knowledge, so then I should proceed further with the discussion, but I got an overview. Thanks for helping out.