* Reconstruct previous epoch without looking at DB: no hindrance to
performance -> removed 15GB
* Store inactivity scores and slashings in MDBX and do not store diffs
for them(they are tiny 700/400 bytes)
* Reduced dumps from every 2048 to 1024 -> Added 5 GB (maybe we should
down it to 768)
* Parallel processing of shuffled sets, 2x performance boost in reading
participation.
* Store balances diffs in a Btree diff matter, see:
https://github.com/ledgerwatch/erigon-documents/blob/master/caplin/design/data-model.md#uint64listuint64vector
"whitelisting" mechanism (list of files - stored in DB) - which
protecting us from downloading new files after upgrade/downgrade was
broken. And seems it became over-complicated with time.
I replacing it by 1 persistent flag inside downloader:
"prohibit_new_downloads.lock"
Erigon will turn downloader into this mode after
downloading/verification of first snapshots.
```
//Corner cases:
// - Erigon generated file X with hash H1. User upgraded Erigon. New version has preverified file X with hash H2. Must ignore H2 (don't send to Downloader)
// - Erigon "download once": means restart/upgrade/downgrade must not download files (and will be fast)
// - After "download once" - Erigon will produce and seed new files
```
------
`downloader --seedbox` is never "prohibit new downloads"
* Most of the PR changed files are extra and slightly more complicated
unit tests.
* Fixed Eth1DataVotes not inheriting genesis
* Fixed Attestations simulation using wrong slot when reconstructing
partecipation
* Fixed Copy() operation on BeaconState on Eth1DataVotes
* Used correct ListSSZ type for Eth1DataVotes and HistoricalSummaries
* Fixed wrong []uint64 deltas on empty slots
What does this PR do:
* Optional Backfilling and Caplin Archive Node
* Create antiquary for historical states
* Fixed gaps of chain gap related to the Head of the chain and anchor of
the chain.
* Added basic reader object to Read the Historical state
adds a two indexes to the validators cache
creates beaconhttp package with many utilities for beacon http endpoint
(future support for ssz is baked in)
started on some validator endpoints
This PR is to add the request rate limiter.
The solution is to count the request number for each peer for each
minute, if the peer exceeds the limit, block the requests for a
specified time.
Current limits:
- Request limited to `5000` requests per minute for each handler.
- Penalty blockage time `1-minute`
This fixes an issue where the mumbai testnet node struggle to find
peers. Before this fix in general test peer numbers are typically around
20 in total between eth66, eth67 and eth68. For new peers some can
struggle to find even a single peer after days of operation.
These are the numbers after 12 hours or running on a node which
previously could not find any peers: eth66=13, eth67=76, eth68=91.
The root cause of this issue is the following:
- A significant number of mumbai peers around the boot node return
network ids which are different from those currently available in the
DHT
- The available nodes are all consequently busy and return 'too many
peers' for long periods
These issues case a significant number of discovery timeouts, some of
the queries will never receive a response.
This causes the discovery read loop to enter a channel deadlock - which
means that no responses are processed, nor timeouts fired. This causes
the discovery process in the node to stop. From then on it just
re-requests handshakes from a relatively small number of peers.
This check in fixes this situation with the following changes:
- Remove the deadlock by running the timer in a separate go-routine so
it can run independently of the main request processing.
- Allow the discovery process matcher to match on port if no id match
can be established on initial ping. This allows subsequent node
validation to proceed and if the node proves to be valid via the
remainder of the look-up and handshake process it us used as a valid
peer.
- Completely unsolicited responses, i.e. those which come from a
completely unknown ip:port combination continue to be ignored.
-
Consensus Specification Tests takes less than 8 minutes so I think they
can be in a PR's own CI for whenever it is ready. for reference it is
less than make test
