6318705607
Add keys
2021-05-03 17:18:54 -07:00
b948a18841
Key rotation
2021-05-03 17:18:54 -07:00
b2778f34f5
Rotate keys
2021-05-03 17:18:54 -07:00
7cea2c4466
validates gossip addresses before sending pull-requests
...
IP addresses need to be validated before sending packets to them.
This commit, sends a ping packet to nodes before any pull requests.
Pull requests are then only sent to the nodes which have responded with
the correct hash of their respective ping packet.
2021-05-03 18:21:06 +00:00
2231017b35
uses Mutex instead of RwLock for ping_cache
2021-05-03 18:21:06 +00:00
a698e34744
patches local pending push messages processing ( #16833 )
...
process_push_messages writes local pending push messages to the crds
table, but it discards the return value:
https://github.com/solana-labs/solana/blob/cf779c63c/core/src/crds_gossip.rs#L96-L102
In order to exclude outdated values from the next pull-request, we need
to record the hash of values purged/overridden by the local push
messages, otherwise pull-responses will return outdated values back to
the node:
https://github.com/solana-labs/solana/blob/c1829dd00/core/src/crds_gossip_pull.rs#L447-L452
Additionally, gossip packets arrive and are processed out of order. So,
local pending push messages should be flushed *before* generating bloom
filters for pull-requests, preventing pull-responses returning the same
values back to the node itself. This requires flipping order of
generating pull and push messages:
https://github.com/solana-labs/solana/blob/cf779c63c/core/src/cluster_info.rs#L1757-L1762
Both above bugs cause redundant traffic and bandwidth waste in gossip
pull-responses.
2021-05-03 16:00:17 +00:00
541aa5ad85
tests: lamports -> lamports() ( #16982 )
2021-05-03 10:45:54 -05:00
8e561354d5
Improve readability of vote lockout processing ( #16987 )
...
* Improve readability of vote lockout processing
* clippy
* simplify comment
* feedback
2021-05-02 08:36:06 +00:00
5981399612
Distinguish max replayed and max observed vote ( #16936 )
2021-04-29 14:43:28 -07:00
542d88929f
Add getBlockProduction RPC method
2021-04-28 20:02:54 -07:00
b5d30846d6
Retry latest vote if expired ( #16735 )
2021-04-28 11:46:16 -07:00
25054bfd35
retains peer's contact-info when making pull requests ( #16715 )
...
ClusterInfo::new_pull_requests has to lookup contact-infos:
https://github.com/solana-labs/solana/blob/a1ef2bd74/core/src/cluster_info.rs#L1663-L1673
when it was already available when making pull requests:
https://github.com/solana-labs/solana/blob/a1ef2bd74/core/src/crds_gossip_pull.rs#L232
2021-04-28 13:19:12 +00:00
1ac2a8cfa5
removes delayed crds inserts when upserting gossip table ( #16806 )
...
It is crucial that VersionedCrdsValue::insert_timestamp does not go
backward in time:
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/crds.rs#L67-L79
Otherwise methods such as get_votes and get_epoch_slots_since will
break, which will break their downstream flow, including vote-listener
and optimistic confirmation:
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/cluster_info.rs#L1197-L1215
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/cluster_info.rs#L1274-L1298
For that, Crds::new_versioned is intended to be called "atomically" with
Crds::insert_verioned (as the comment already says so):
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/crds.rs#L126-L129
However, currently this is violated in the code. For example,
filter_pull_responses creates VersionedCrdsValues (with the current
timestamp), then acquires an exclusive lock on gossip, then
process_pull_responses writes those values to the crds table:
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/cluster_info.rs#L2375-L2392
Depending on the workload and lock contention, the insert_timestamps may
well be in the past when these values finally are inserted into gossip.
To avoid such scenarios, this commit:
* removes Crds::new_versioned and Crd::insert_versioned.
* makes VersionedCrdsValue constructor private, only invoked in
Crds::insert, so that insert_timestamp is populated right before
insert.
This will improve insert_timestamp monotonicity as long as Crds::insert
is not called with a stalled timestamp. Following commits may further
improve this by calling timestamp() inside Crds::insert, and/or
switching to std::time::Instant which guarantees monotonicity.
2021-04-28 11:56:13 +00:00
b17d5eeaee
moves cluster-info metrics to a separate module ( #16883 )
2021-04-28 02:04:49 +00:00
b468ead1b1
uses current timestamp when flushing local pending push queue ( #16808 )
...
local_message_pending_push_queue is recording timestamps at the time the
value is created, and uses that when the pending values are flushed:
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/cluster_info.rs#L321
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/crds_gossip.rs#L96-L102
which is then used as the insert_timestamp when inserting values in the
crds table:
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/crds_gossip_push.rs#L183
The flushing may happen 100ms after the values are created (or even
later if there is a lock contention). This will cause non-monotone
insert_timestamps in the crds table (where time goes backward),
hindering the usability of insert_timestamps for other computations.
For example both ClusterInfo::get_votes and get_epoch_slots_since rely
on monotone insert_timestamps when values are inserted into the table:
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/cluster_info.rs#L1197-L1215
https://github.com/solana-labs/solana/blob/ec37a843a/core/src/cluster_info.rs#L1274-L1298
This commit removes timestamps from local_message_pending_push_queue and
uses current timestamp when flushing the queue.
2021-04-28 00:15:11 +00:00
bc31378797
Trim extra shred bytes in blockstore ( #16602 )
...
Strip the zero-padding off of data shreds before insertion into blockstore
Co-authored-by: Stephen Akridge <sakridge@gmail.com>
Co-authored-by: Nathan Hawkins <utsl@utsl.org>
2021-04-27 17:40:41 -05:00
3b8d6b59fb
records hash of values purged by expired pull-responses ( #16800 )
...
process_pull_responses should record hash of values purged by expired
responses (as well as unexpired ones):
https://github.com/solana-labs/solana/blob/c1829dd00/core/src/crds_gossip_pull.rs#L385-L387
otherwise, these values are not excluded from following pull-requests
(from likely different nodes):
https://github.com/solana-labs/solana/blob/c1829dd00/core/src/crds_gossip_pull.rs#L447-L452
and would waste bandwidth should they be included in subsequent
pull-responses.
2021-04-27 12:06:49 +00:00
0f3ac51cf1
limits to data_header.size when combining shreds' payloads ( #16708 )
...
Shredder::deshred is ignoring data_header.size when combining shreds' payloads:
https://github.com/solana-labs/solana/blob/37b8587d4/ledger/src/shred.rs#L940-L961
Also adding more sanity checks on the alignment of data shreds indices.
2021-04-27 12:04:44 +00:00
59fc33635a
Add getVoteAccounts RPC method parameter to restrict results to a single vote account
2021-04-27 04:27:15 +00:00
9706512115
removes old runtime feature gates in gossip and turbine ( #16633 )
2021-04-26 17:12:02 +00:00
ca14c18998
owner -> owner() ( #16782 )
2021-04-23 22:49:47 +00:00
63436cc2bf
Disable flaky test_poh_service ( #16772 )
2021-04-23 12:14:11 -05:00
2c82f2154d
retains crds values if the origin is still active ( #16576 )
...
Local timestamps are updated for records associated with a pubkey if the
origin is still active:
https://github.com/solana-labs/solana/blob/c8ed14c64/core/src/crds.rs#L301-L311
However this is done inconsistently on some gossip paths (pull requests
and pull responses) but not all (e.g. push messages). Additionally
update_record_timestamp is inefficient since there can be ~800 values
associated with each pubkey.
This commit updates records timestamps only on contact-infos; and,
instead utilizes origin's timestamp when purging old values.
2021-04-23 15:14:49 +00:00
03194145c0
removes first_coding_index from erasure recovery code ( #16646 )
...
first_coding_index is the same as the set_index and is so redundant:
https://github.com/solana-labs/solana/blob/37b8587d4/ledger/src/blockstore_meta.rs#L49-L60
2021-04-23 12:00:37 +00:00
75b8434b76
Add TPU client for sending txs to the current leader tpu port ( #16736 )
...
* Add TPU client for sending txs to the current leader tpu port
* Update tpu_client.rs
2021-04-23 09:35:12 +08:00
636b5987af
Update getLeaderSchedule options ( #16749 )
2021-04-22 19:27:30 +00:00
6004c0abf5
getLeaderSchedule now supports filtered results based on validator identity
2021-04-21 17:59:26 -07:00
91b6888e15
verify_pubkey() now takes a ref
2021-04-21 14:43:49 -07:00
4c94f8933f
Ingest votes from gossip into fork choice ( #16560 )
2021-04-21 14:40:35 -07:00
a1ef2bd74d
Ignore flaky test_pull_request_time_pruning
2021-04-21 12:07:36 -07:00
37b8587d4e
expands number of erasure coding shreds in the last batch in slots ( #16484 )
...
Number of parity coding shreds is always less than the number of data
shreds in FEC blocks:
https://github.com/solana-labs/solana/blob/6907a2366/ledger/src/shred.rs#L719
Data shreds are batched in chunks of 32 shreds each:
https://github.com/solana-labs/solana/blob/6907a2366/ledger/src/shred.rs#L714
However the very last batch of data shreds in a slot can be small, in
which case the loss rate can be exacerbated.
This commit expands the number of coding shreds in the last FEC block in
slots to: 64 - number of data shreds; so that FEC blocks are always 64
data and parity coding shreds each.
As a consequence of this, the last FEC block has more parity coding
shreds than data shreds. So for some shred indices we will have a coding
shred but no data shreds. This should not cause any kind of overlapping
FEC blocks as in:
https://github.com/solana-labs/solana/pull/10095
since this is done only for the very last batch in a slot, and the next
slot will reset the shred index.
2021-04-21 12:47:50 +00:00
0924c2d070
Add port and gossip options to solana-test-validator ( #16696 )
2021-04-21 02:40:52 +00:00
34addee882
getVoteAccounts: Limit the length of the `epoch_credits` array
2021-04-20 14:42:28 -07:00
8e69dd42c1
Add non-default repair nonce values ( #16512 )
...
* Track outstanding nonces in repair
* Rework outstanding requests to use lru cache and randomize nonces
Co-authored-by: Carl <carl@solana.com>
2021-04-20 09:37:33 -07:00
bc90e04e64
uses current local timestamp when recording purged values
...
CrdsGossipPull.purged_values is meant to record recently purged values
so that they are excluded from imminent pull requests, until the entire
cluster have synced to the updated value:
https://github.com/solana-labs/solana/blob/c826cddbb/core/src/crds_gossip_pull.rs#L449-L454
However, VersionedCrdsValue.local_timestamp represents the local time
when the value was last updated, and given that crds values may have
different timeouts based on stake, it does not necessarily represent how
recently the value was purged:
https://github.com/solana-labs/solana/blob/c826cddbb/core/src/crds.rs#L75-L76
As such, recording current local timestamp when purging values is more
appropriate. Additionally, purge_purged assumes that the purge_values is
sorted in timestamps when draining the old ones; which is not true if
those timestamps are VersionedCrdsValue.local_timestamp:
https://github.com/solana-labs/solana/blob/c826cddbb/core/src/crds_gossip_pull.rs#L563-L571
2021-04-20 11:21:00 +00:00
a7e65c0034
RPC: use finalized as default pubsub commitment level ( #16659 )
...
* RPC: use finalized as default pubsub commitment level
* update docs
* Fix tests
2021-04-20 08:19:54 +00:00
c8b474cd0b
Send votes to next leader's TPU instead of our TPU
2021-04-20 00:38:21 -07:00
4aa753ff01
rename threads: 15 char limit ( #16625 )
2021-04-19 12:16:58 -05:00
b06e93fe5b
Increase test timeout
2021-04-18 20:55:02 -07:00
a911ae00ba
clippy
2021-04-18 20:55:02 -07:00
e405747409
Revert "Add limit and shrink policy for recycler ( #15320 )"
...
This reverts commit c2e8814dce
2021-04-18 19:29:24 +00:00
6907a2366e
Remove unnecessary clone
2021-04-17 10:23:13 -07:00
bb24318ef0
Document shreds ( #16514 )
...
No functionality changes from this commit
2021-04-16 14:04:46 -05:00
974e6dd2c1
Deprecate "confirmed" RpcClient methods ( #16520 )
...
* Remove obsolete client methods
* Deprecate GetConfirmed client methods
* Rename Confirmed config structs, with appropriate deprecation
* Fixup client apps
* Map RpcRequest to deprecated when targeting older nodes
2021-04-15 17:00:14 -06:00
7dfb51c0b4
Cli: move airdrop to rpc requests ( #16557 )
...
* Add recent_blockhash to requestAirdrop
* Move tx confirmation to separate method
* Add RpcClient airdrop methods
* Request cli airdrop via RpcClient
* Pass optional faucet_addr into TestValidator and fix tests
* Update client/src/rpc_client.rs
Co-authored-by: Michael Vines <mvines@gmail.com>
Co-authored-by: Michael Vines <mvines@gmail.com>
2021-04-15 06:25:23 +00:00
d92721aab9
uses timeouts based on stake for filtering pull responses ( #16549 )
...
filter_pull_responses is using default timeout when discarding pull
responses (except for ContactInfo):
https://github.com/solana-labs/solana/blob/f804ce63c/core/src/crds_gossip_pull.rs#L349-L350
But purging code uses timeouts based on stake:
https://github.com/solana-labs/solana/blob/f804ce63c/core/src/cluster_info.rs#L1867-L1870
So the crds value will not be purged from the sender's table and will be
sent again over the next pull request.
2021-04-14 20:18:00 +00:00
f35a6a8be0
prioritizes contact-infos in pull responses ( #16541 )
...
Expired crds values where the contact-info does not exist are wasted:
https://github.com/solana-labs/solana/blob/f804ce63c/core/src/crds_gossip_pull.rs#L353-L378
and then are sent again over the next pull-request.
Also, the stake of the first response (which can be anything) is used to
weight all pull-responses to a node, while the rest of responses can
have different stake.
https://github.com/solana-labs/solana/blob/f804ce63c/core/src/cluster_info.rs#L2231
2021-04-14 18:45:20 +00:00
f0c150cfb9
Fix channel panic in tests ( #16503 )
...
* Fix channel panic
* Add exit signal to PohRecorder because Crossbeam doesnt drop objects inside dropped channel
2021-04-14 12:07:21 -05:00
f641429056
Remove blake3 from bpf program dependencies ( #16506 )
2021-04-13 10:06:05 +00:00
37afa00ffb
Rpc: deprecate getConfirmed endpoints ( #16502 )
...
* Deprecate getConfirmed methods in rpc
* Add new methods to docs
* Move deprecated rpc methods to separate docs section
* Add note to docs about removal timing
2021-04-13 01:50:15 -06:00
publish-docs.sh
behzad nouri
Jeff Washington (jwash)
Justin Starry
carllin
Michael Vines
steviez
Tyera Eulberg
sakridge