To measure whether the stashing of parsoid base html for VE is working properly, we cen define an SLO based on the probability of stash read misses, as a function of the time since the stash entry was written. In order to measure compliance with the SLO, we'll want to track the following:

• Stash hits vs. stash miss. Misses should be minimal, since the stash should have near perfect persistance - unless the TTL was exceeded, or the clinet submitted a bad stash key.
• The time between stash write and stash read.
• The time between stash write and stash miss. We need to reply on a timestamp supplied by the client, since for a stash miss, we don't know when and if the data was written.

Since we rely on client supplied information for the metrics, we may want to have them HMAC signed, to avoid tempering.