After running into some scaiing problems with our Elasticsearch cluster (running as part of an ELK stack), I read up on refreshes, and in particular, the refresh interval. I set it from 1s to 30s (which should be totally acceptable for our needs), and performance improved dramatically, which was wonderful.
That is, until the day ended and a new set of indices were created, which had the default value of 1s again.
The command I originally issued to set the interval to 30s for all indices:
curl -XPUT hostname:9200/_settings -d '{
"index": {
"refresh_interval": "30s"
}
}'
Is there some different command to have it take effect on a cluster-wide basis? Is this something I can do with index templates? If so, is that the only way to do so?