If you happen to run the installation wizard after you've already configured your node, when reaching the "Datastore size" step, the wizard will load your current datastore size parameter.

However, when you choose a value as datastore size, the wizard will then apply the following operations:

• compute a clientCacheSize, of 10% or 200 MiB (whichever is smaller)
• compute a slashdotCacheSize of up to 10%
• compute storeSize = [your chosen size] - clientCacheSize - slashdotCacheSize, and use this as the datastore size.

So, the actual value for datastore size will be different from the chosen value. In particular, if you choose "current size", in the end the datastore will get reduced, leading to a datastore resize (which may take a long time and will lead to its partial purge).

Example

For instance, if I choose 10 GiB, then in the logs I'll see:

Setting datastore size to 9016MiB
Setting client cache size to 200MiB
Setting slashdot/ULPR/recent requests cache size to 1024MiB

And if I go back to the datastore size wizard step, "current size" will then show 8.8 GiB (the equivalent of 9016 MiB), instead of 10 GiB (what I chose).

Proposed fix

In this PR, I compute storeSize + clientCacheSize + slashdotCacheSize, and I then use this to fill in the "current size" option, instead of just storeSize.

Although this is a bit inaccurate as far as what the current size of the datastore really is, it seems to me that it leads to a more coherent behavior.