I increased the Data RAM quota in the settings tab. An hour has passed after I made the change. The cluster does not seem to want to use the other RAM although only ~20% of documents are in RAM. Here’s a screenshot of the overview.
What can I do so that the cluster uses the unallocated RAM?
There’s no “speculative paging” of items into free memory currently - until a non-resident item is operated on (read or modified) it won’t be fetched from disk and cached in memory.
Thanks @drigby! Your answer makes perfect sense. Would you also know why the values “low water mark” and “high water mark” did not change yet. Is this also expected behaviour?
High and Low watermark are per-bucket. Have you also increased the individual buckets’ RAM quota after increasing the per-server quota?
That helped! First I edited the Data RAM quota in Setting - Cluster - Cluster RAM Quota. Now I edited the bucket and the low and high water mark values went up immediately. Thanks for helping!