We have a system that from one side writes information, and on the other sides reads the data.
I synchronize the data based on a timestamp. I read data from a point in time, arrange it according to the timestamp and continue reading from the last timestamp I got.
I noticed that if I always read the latest data, i miss some data which is written out of order, so instead I added a delay of 1 minute, and now everything is fine.
My question is, what is the minimal delay I should use? In other words, what is the "settling time" of written records, and does it depend on the throughput of data?
Once you have the response of elasticsearch, it can take up to 1s before the document is actually searchable. This is called refresh.
You can also make your index query wait for the next refresh to happen in order to send back the index response.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.