I have problem with jumbochunk, but not with situation that one of my chunk was marked as jumbo but with situation where chunk should be marked as jumbo but was not marked.
We have situation the same like in https://jira.mongodb.org/browse/SERVER-26688 and because exists issue in Mongo https://jira.mongodb.org/browse/SERVER-26531 chunk was not marked as jumbo.
Why this is problem for us - for each update on collection mongo call splitVector query and try split chunk. In this collection we have 26 million of documents and splitVector function overloads one of our node. Each update calls next splitVector query and node not does anything more but only try split this chunk.
In correct situation this chunk should have jumbo flag and mongo will be not call splitVector query.
I know correct solution we should re-shard collection with different shard key.
But I want ask you community, Do you know any workaround for issue https://jira.mongodb.org/browse/SERVER-26531, that workaround should preventing situation where mongo clears inadvertently jumbo flag. For now, I want to have chunk marked as jumbo, like should be.
Best regards, Arek and Thanks in advance.