How do I use retry_on_conflict to resolve error "ConflictError 409" with update()?

[hoIIer]

I noticed that if I send too many requests in a short period, my elasticsearch update code throws a 409 ConflictError. Reading a bit into it, I noticed there is a param for update() to set the number of retry attempts, however when I tried using it, I get another error:

RequestError(400, 'action_request_validation_exception', 'Validation Failed: 1: compare and write operations can not be retried;')

How can I handle 409 ConflictError during frequent updates and/or use the retry_on_conflict param?

    thing = Thing.get(id=thing_id, ignore=404)

    if thing:
            return post.update(
                script=update_func,
                refresh=True,
                retry_on_conflict=ES_UPDATE_RETRY_COUNT,
                now=arrow.utcnow().float_timestamp,
            )

ConflictError: ConflictError(409, 'version_conflict_engine_exception', '[TtWP-la58FPHbw]: version conflict, required seqNo [8], primary term [1]. current document has seqNo [9] and primary term [1]')
(2 additional frame(s) were not displaye

[hoIIer]

using elasticsearch-py directly seems to work...

            return post._get_connection().update(
                Post.Index().name,
                post_id,
                body,
                refresh=True,
                retry_on_conflict=10,
            )

[549982170]

I have the same problem.

[smurf-U]

I have the same problem.

Any update?

after adding retry_on_conflict I'm getting below one RequestError(400, 'action_request_validation_exception', 'Validation Failed: 1: compare and write operations can not be retried;')

Any soulution?

[hoIIer]

@smurf-U I ended up using above solution

[smurf-U]

Hi @burritoIand
Thanks for reply.
Then how did you handle this scenario?

[hoIIer]

@smurf-U

post.update(
                script=update_func,
                refresh=True,
                retry_on_conflict=ES_UPDATE_RETRY_COUNT,
                now=arrow.utcnow().float_timestamp,
            )

[ryonakamura]

This error seems to occur when adding if_seq_no and if_primary_term to the update arguments.

https://github.com/elastic/elasticsearch-dsl-py/blob/7.3.0/elasticsearch_dsl/document.py#L407

Here's a temporary solution:

del post.meta['seq_no']
del post.meta['primary_term']
post.update(
    retry_on_conflict=10,
    date=datetime.utcnow(),
)