Hi,
My system is on v 7.4.6.
I have this strange issue where i get duplicate entries in index.
Only replace command is used to update documents.
I have read about this issue with big ids but no big ids are used (max is 182711754).
Setup:
Manticore Search cluster with 3 nodes
seems like a bug.
If you have reproduced example it could be better to create ticket at Github there to put all data that recreates this issue locally.
Otherwise you have to truncate your RT table and repopulates it with the data.
I have only reproduced this issue in my production environment where i have around 7 million records. I have replace rate of 20 docs/sec and delete rate bellow 1 doc/sec. Whole dump is around 7GB. What data should i provide?
I have already tried to truncate table several times but issue persists.
you could try to minimize data, try to catch the case there the same id got duplicated then upload the index files with the requests stream into write-only S3 storage as explained Reporting bugs | Manticore Search Manual
Do you post replace and delete at the same node or multiple nodes used to populate cluster with data?
Do you got duplicates for the same id or every try you got different duplicates?
Do you got the same duplicates and across nodes at the cluster or different nodes has different duplicates?
Ok, thanx. Will do on monday.
