mainnet fork upgrade14 test with time measurement

[aj-agoric]

What is the Problem Being Solved?

With concern regarding mainnet performance of a few changes in this release —

release notes say IAVL upgrade took 15min on Osmosis
we do not have data on how rapidly KREAd core-proposal will run.

We aim to produce a run of upgrade14 on a mainfork to get produce accurate time estimates

Description of the Design

Security Considerations

Scaling Considerations

Test Plan

Upgrade Considerations

[gibson042]

ollinet mainfork upgrade with iavl-disable-fastnode = false

Time	Duration	Concluding Event
20:40 (approx)	-	start new software
21:55:39	1h15m39s	message:starting node with ABCI Tendermint in-process (no actual logs before that)
21:56:37	58s	Launching SwingSet Kernel
21:57:34	57s	Launched SwingSet kernel
21:57:47	13s	block-manager: block 13718557 begin
21:57:48	1s	vat: v1: coreProposal: upgradeWalletFactory
21:58:52	1m4s	kernel: vat v43 upgraded from incarnation 1 to 2
22:01:02	2m10s	vat: v1: Successfully upgraded WalletFactory
22:04:54	3m52s	vat: v1: coreProposal: upgradeZcf
22:04:57	3s	kernel: vat v9 upgraded from incarnation 0 to 1
22:09:10	4m13s	vat: v1: coreProposal: repairKread
22:09:52	42s	block 13718557 commit

[mhofman]

State sync snapshot creation timing on mainfork

event time	event name
17:23:02	block 13731000 commit
17:23:02	snapshot initiate
18:04	swing-store export complete
18:29:09	swing-store export retrieve
18:51:53	swing-store artifacts added to state-sync snapshot
18:52:03	state-sync snapshot complete

Snapshot complete at block 13731991

[mhofman]

The mainfork snapshot was taken from export of agoric-3 at height 13512995. As of today, agoric-3 is at height 13986253.

[mhofman]

We did some testing on a snapshot from about January which informed the expected upgrade time for mainnet and emerynet. While we could do a new test with a more recent snapshot, it's unlikely we'll learn anything more worth the effort.