Fix RPC_process opens the store issues

List of issues associated to MR !9490 (merged)

Failing tests

• 'Alpha: storage snapshot export and import' (@vect0r)
  • flaky test (!9490 (merged))
• 'Alpha: forge block with wrong payload' (@vect0r)
  • flaky test: op is sometimes missing from block (!9815 (merged))
• 'manually forked migration blocks from nairobi to alpha' (@vect0r)
  • flaky test: op is sometimes missing from block (!9815 (merged))
• 'Alpha: forge block with wrong payload' (@dianegalloiswong )
  • flaky test (even or master): precheck failure
  • due to the injection of on a block in the ring instead of the node
  • presumably caused by a port clash
• 'Oxford: node synchronization (archive / archive)' (@vect0r)
  • flaky test:
  • cannot open store while node is actually merging the store
  • cannot rename error caused by the Index.init ~readonly:true during a merge
• 'Alpha: VDF daemon' (@remyzorg) !9922 (merged)
  • test failing [error] File "none", line 0, characters -1-5: Assertion failed
• 'amendment: alpha -> injected_test (losers: nairobi)' (@remyzorg)
  • test failing

  • [09:53:40.556] [node1]   Raised at Tezos_event_logging__Internal_event.Section.is_prefix.(fun) in file "src/lib_event_logging/internal_event.ml", line 158, characters 19-34```

• 'Nairobi: Manager_restriction_propagation' (@dianegalloiswong)
  • flaky test: Fitness too low
• Double baking flakiness: Cannot find protocol X
  • dune exec tezt/tests/main.exe -- --file tezt/tests/double_bake.ml --title 'Alpha: double baking with accuser' --loop-count 1000 --test-timeout 120 --verbose |& tee tezt.log
  • It seems that the protocol table is updated after a new head is promoted, on the shell part. However, the rpc process may receive the notification of an applied block in the mean time. As the protocol table is not yet updated, et fails to get the block

Various Todos

• complementary list of non-blocking vs. blocking RPCs (@remyzorg)
  • Is that really necessary? A bit complicated and not that useful
• Stop calling Store.sync while the node is bootstrapping (@vect0r)
  • Induces bad perfs
• Check that mempool config RPCs are well forwarded (@dianegalloiswong)
  • A test should already tackle that
• testchain RPC must be forwarded to the node in not tackled
  • confirm that on /monitor/head
• Bench the branch
  • make sure that the Store.sync is not slowing down too much the injection of pre-/attestation