(cloud) big performance fixes

things-to-search-for.txt

@@ -1,15 +1,9 @@
 or-join syntax changed?
-Make sure no history on ledger
+regex no longer supported as argument for query. Check vendor datomic for how to do it right.
 it looks like there are a bbunch of orrphaned customizations for accounts, breaking indexes
 ezcater graphql needs search index too
 make sure that temporary ids are set on all new things when using upsert-entity
 Wrap tests around every api call
-Fix tests
-Make a different solution for running balance cache
-* Store the running balance on the line
-* Make ledger changes mark as dirty
-* Recompute the cache periodically, somewhat randomly to avoid races
-* Have a background job recompute all at night.
 
 Fix searching
 * indexing should happen more regularly, and just look for changes since last time it was run
@@ -17,9 +11,27 @@ Fix searching
 Running Balance Cache
 * Add tests for upsert-ledger
 * try again to see if we can get upsert-ledger into the same transaction, making it all or nothing
-* make rest of rebuilding the cache use new index
-* ensure somehow that the index is always right
-
+* Make a new way to reset the entire cache for a client
 
 Address memory
 * JVM settings now and in prod
+
+Release steps:
+Stop prod
+Make database snapshot
+Create new database for prod-cloud (just called prod)
+Restore database
+Transact new schema
+(reset-client+account+location+date)
+(force-rebuild-running-balance-cache)
+Merge branch into master
+Rename prod-cloud to prod everywhere
+Release again
+
+Sanity checks later:
+* Run query 
+
+
+Future improvements:
+Make reports just be based on running-balances
+