0c0384ec32
revises turbine peers shuffling order ( #20480 )
...
Turbine randomly shuffles cluster nodes on a broadcast tree for each
shred. This requires knowing the stakes and nodes' contact-infos (from
gossip).
However gossip is subject to partitioning and propogation delays.
Additionally unstaked nodes may join and leave the cluster at any
moment, changing the cluster view from one node to another.
This commit:
* Always arranges the unstaked nodes at the bottom of turbine broadcast
tree.
* Staked nodes are always included regardless of if their contact-info
is available in gossip or not.
* Uses the unbiased WeightedShuffle construct for shuffling nodes.
2021-10-14 15:09:36 +00:00
6d9818b8e4
skips retransmit for shreds with unknown slot leader ( #19472 )
...
Shreds' signatures should be verified before they reach retransmit
stage, and if the leader is unknown they should fail signature check.
Therefore retransmit-stage can as well expect to know who the slot
leader is and otherwise just skip the shred.
Blockstore checking signature of recovered shreds before sending them to
retransmit stage:
https://github.com/solana-labs/solana/blob/4305d4b7b/ledger/src/blockstore.rs#L884-L930
Shred signature verifier:
https://github.com/solana-labs/solana/blob/4305d4b7b/core/src/sigverify_shreds.rs#L41-L57
https://github.com/solana-labs/solana/blob/4305d4b7b/ledger/src/sigverify_shreds.rs#L105
2021-09-01 15:44:26 +00:00
1deb4add81
removes Slot from TransmitShreds ( #19327 )
...
An earlier version of the code was funneling through stakes along with
shreds to broadcast:
https://github.com/solana-labs/solana/blob/b67ffab37/core/src/broadcast_stage.rs#L127
This was changed to only slots as stakes computation was pushed further
down the pipeline in:
https://github.com/solana-labs/solana/pull/18971
However shreds themselves embody which slot they belong to. So pairing
them with slot is redundant and adds rooms for bugs should they become
inconsistent.
2021-08-20 13:48:33 +00:00
aa32738dd5
uses cluster-nodes cache in broadcast-stage
...
* Current caching mechanism does not update cluster-nodes when the epoch
(and so epoch staked nodes) changes:
https://github.com/solana-labs/solana/blob/19bd30262/core/src/broadcast_stage/standard_broadcast_run.rs#L332-L344
* Additionally, the cache update has a concurrency bug in which the
thread which does compare_and_swap may be blocked when it tries to
obtain the write-lock on cache, while other threads will keep running
ahead with the outdated cache (since the atomic timestamp is already
updated).
In the new ClusterNodesCache, entries are keyed by epoch, and so if
epoch changes cluster-nodes will be recalculated. The time-to-live
eviction policy is also encapsulated and rigidly enforced.
2021-08-05 21:47:33 +00:00
44b11154ca
sends slots (instead of stakes) through broadcast flow
...
Current broadcast code is computing stakes for each slot before sending
them down the channel:
https://github.com/solana-labs/solana/blob/049fb0417/core/src/broadcast_stage/standard_broadcast_run.rs#L208-L228
https://github.com/solana-labs/solana/blob/0cf52e206/core/src/broadcast_stage.rs#L342-L349
Since the stakes are a function of epoch the slot belongs to (and so
does not necessarily change from one slot to another), forwarding the
slot itself would allow better caching downstream.
In addition we need to invalidate the cache if the epoch changes (which
the current code does not do), and that requires to know which slot (and
so epoch) current broadcasted shreds belong to:
https://github.com/solana-labs/solana/blob/19bd30262/core/src/broadcast_stage/standard_broadcast_run.rs#L332-L344
2021-08-05 21:47:33 +00:00
14361906ca
for all tests, bank::new -> bank::new_for_tests ( #19064 )
2021-08-05 08:42:38 -05:00
049fb0417f
allows sendmmsg api taking owned values (as well as references) ( #18999 )
...
Current signature of api in sendmmsg requires a slice of inner
references:
https://github.com/solana-labs/solana/blob/fe1ee4980/streamer/src/sendmmsg.rs#L130-L152
That forces the call-site to convert owned values to references even
though doing so is redundant and adds an extra level of indirection:
https://github.com/solana-labs/solana/blob/fe1ee4980/core/src/repair_service.rs#L291
This commit expands the api using AsRef and Borrow traits to allow
calling the method with owned values (as well as references like
before).
2021-07-30 20:58:49 +00:00
84e78316b1
Write helper for multithread update ( #18808 )
2021-07-29 03:16:36 +02:00
d2d5f36a3c
adds validator flag to allow private ip addresses ( #18850 )
2021-07-23 15:25:03 +00:00
e316586516
excludes private ip addresses
2021-07-16 20:05:48 -06:00
ae5ad5cf9b
sendmmsg cleanup #18589
...
Rationalize usage of sendmmsg(2). Skip packets which failed to send and track failures.
2021-07-16 14:36:49 -07:00
7f2254225e
Move entry/poh to own crate to speed up poh bench build ( #18225 )
2021-07-14 14:16:29 +02:00
175083c4c1
Add updated duplicate broadcast test ( #18506 )
2021-07-10 22:22:07 -07:00
a86ced0bac
generate deterministic seeds for shreds ( #17950 )
...
* generate shred seed from leader pubkey
* clippy
* clippy
* review
* review 2
* fmt
* review
* check
* review
* cleanup
* fmt
2021-07-07 08:21:12 -07:00
04787be8b1
encapsulates turbine peers computations of broadcast & retransmit stages ( #18238 )
...
Broadcast stage and retransmit stage should arrange nodes on turbine
broadcast tree in exactly same order. Additionally any changes to this
ordering (e.g. updating how unstaked nodes are handled) requires feature
gating to keep the cluster in sync.
Current implementation is scattered out over several public methods and
exposes too much of implementation details (e.g. usize indices into
peers vector) which makes code changes and checking for feature
activations more difficult.
This commit encapsulates turbine peer computations into a new struct,
and only exposes two public methods, get_broadcast_peer and
get_retransmit_peers, for call-sites.
2021-07-07 00:35:25 +00:00
b6792a3328
Add ability to change the validator identity at runtime
2021-07-01 17:50:04 -07:00
84b9de8c18
Shredder no longer holds a keypair
2021-06-21 21:29:52 -07:00
4a12c715a3
Drop Error suffix from enum values to avoid the enum_variant_names clippy lint
2021-06-18 23:02:13 +00:00
6514096a67
chore: cargo +nightly clippy --fix -Z unstable-options
2021-06-18 10:42:46 -07:00
050bb5446d
Add local cluster tests that broadcast duplicate slots ( #13995 )
...
* Add duplicate node local cluster test
* fix clippy
* remove dupe test
2021-06-09 15:01:48 -07:00
544b3c0d17
Create solana-poh and move remaining rpc modules to solana-rpc ( #17698 )
...
* Create solana-poh crate
* Move BigTableUploadService to solana-ledger
* Add solana-rpc to workspace
* Move dependencies to solana-rpc
* Move remaining rpc modules to solana-rpc
* Single use statement solana-poh
* Single use statement solana-rpc
2021-06-04 09:23:06 -06:00
9a5330b7eb
Move gossip modules into solana-gossip crate ( #17352 )
...
* Move gossip modules to solana-gossip
* Update Protocol abi digest due to move
* Move gossip benches and hook up CI
* Remove unneeded Result entries
* Single use statements
2021-05-26 09:15:46 -06:00
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
570fd3f810
makes turbine peer computation consistent between broadcast and retransmit ( #14910 )
...
get_broadcast_peers is using tvu_peers:
https://github.com/solana-labs/solana/blob/84e52b606/core/src/broadcast_stage.rs#L362-L370
which is potentially inconsistent with retransmit_peers:
https://github.com/solana-labs/solana/blob/84e52b606/core/src/cluster_info.rs#L1332-L1345
Also, the leader does not include its own contact-info when broadcasting
shreds:
https://github.com/solana-labs/solana/blob/84e52b606/core/src/cluster_info.rs#L1324
but on the retransmit side, slot leader is removed only _after_ neighbors and
children are computed:
https://github.com/solana-labs/solana/blob/84e52b606/core/src/retransmit_stage.rs#L383-L384
So the turbine broadcast tree is different between the two stages.
This commit:
* Removes retransmit_peers. Broadcast and retransmit stages will use tvu_peers
consistently.
* Retransmit stage removes slot leader _before_ computing children and
neighbors.
2021-03-24 13:34:48 +00:00
4f82b897bc
buffers data shreds to make larger erasure coded sets ( #15849 )
...
Broadcast stage batches up to 8 entries:
https://github.com/solana-labs/solana/blob/79280b304/core/src/broadcast_stage/broadcast_utils.rs#L26-L29
which will be serialized into some number of shreds and chunked into FEC
sets of at most 32 shreds each:
https://github.com/solana-labs/solana/blob/79280b304/ledger/src/shred.rs#L576-L597
So depending on the size of entries, FEC sets can be small, which may
aggravate loss rate.
For example 16 FEC sets of 2:2 data/code shreds each have higher loss
rate than one 32:32 set.
This commit broadcasts data shreds immediately, but also buffers them
until it has a batch of 32 data shreds, at which point 32 coding shreds
are generated and broadcasted.
2021-03-23 14:52:38 +00:00
5df36aec7d
Pacify clippy
2021-02-19 20:08:41 -08:00
e1021d9f83
removes redundant epoch stakes cache in retransmit ( #14781 )
...
Following d6d76219b
2021-01-24 21:15:09 +00:00
cbffab7850
Upgrade to Rust v1.49.0
2021-01-23 19:16:36 -08:00
f8a4afc7c1
Fix flaky broadcast test ( #14329 )
2020-12-29 12:35:04 -08:00
c693ffaa08
Fix subtraction overflow in metrics ( #14290 )
2020-12-27 16:26:22 -08:00
d6d76219b6
caches staked nodes computed from vote-accounts ( #13929 )
2020-12-17 21:22:50 +00:00
7143aaa89b
Clippy
2020-12-14 08:03:29 -08:00
b4cf968e14
Add back shredding broadcast stats ( #13463 )
2020-11-09 23:04:27 -08:00
d15173ad9d
Address latest nightly clippy lints, but globally disable stable_sort_primitive
2020-08-17 22:36:10 -07:00
2cf719ac2c
Cache tvu peers for broadcast ( #10373 )
2020-06-03 08:24:05 -07:00
97f2bcff69
master: Add nonce to shreds repairs, add shred data size to header ( #10109 )
...
* Add nonce to shreds/repairs
* Add data shred size to header
Co-authored-by: Carl <carl@solana.com>
2020-05-19 12:38:18 -07:00
58ef02f02b
9951 clippy errors in the test suite ( #10030 )
...
automerge
2020-05-15 09:35:43 -07:00
bab3502260
Push down cluster_info lock ( #9594 )
...
* Push down cluster_info lock
* Rework budget decrement
Co-authored-by: Carl <carl@solana.com>
2020-04-21 12:54:45 -07:00
77fb4230d6
Calculate distance between u64 without overflow ( #9592 )
...
* fix overflow
* fixed num_live_peers overflow
2020-04-19 23:05:26 -07:00
7aa4d401f7
Fix broadcast metrics ( #9461 )
...
* Rework broadcast metrics to support multiple threads
* Update dashboards
Co-authored-by: Carl <carl@solana.com>
2020-04-15 15:22:16 -07:00
69f1e487b3
Reduce cluster-info metrics. ( #9465 )
2020-04-14 21:21:58 -07:00
d5a9ee97f2
Add --allow-dead-slots argument to `slot`/`print`/`json` commands ( #9408 )
...
automerge
2020-04-09 20:10:51 -07:00
4677cdb4c2
Optimize broadcast cluster_info critical section ( #9327 )
2020-04-06 17:36:22 -07:00
7b68628e6c
Remove write lock ( #9311 )
...
* Remove write lock
Co-authored-by: Carl <carl@solana.com>
2020-04-05 15:18:45 -07:00
18c1f0dfe9
Remove stub core/src/genesis_utils.rs ( #8999 )
2020-03-21 10:54:40 -07:00
dc1db33ec9
Add Capabilities to Signal BroadcastStage to Retransmit ( #8899 )
2020-03-19 23:35:01 -07:00
ab361a8073
Rename KeypairUtil to Signer ( #8360 )
...
automerge
2020-02-20 13:28:55 -08:00
b5dba77056
Rename blocktree to blockstore ( #7757 )
...
automerge
2020-01-13 13:13:52 -08:00
97589f77f8
Pipeline broadcast socket transmit and blocktree record ( #7481 )
...
automerge
2019-12-16 17:11:18 -08:00
6bfe0fca1f
Add a version field to shreds ( #7023 )
...
* Add a version field to shreds
* Clippy
* Fix Chacha Golden
* Fix shredder bench compile
* Fix blocktree bench compile
2019-11-18 18:05:02 -08:00
behzad nouri
Jeff Washington (jwash)
sakridge
Jeff Biseda
carllin
Michael Vines
Alexander Meißner
Justin Starry
Tyera Eulberg
carllin
Kristofer Peterson
anatoly yakovenko
Greg Fitzgerald
Sagar Dhawan