The light client data structures were changed to accommodate additional
fields in future forks (e.g., to also hold execution data).
There is a minor change to the JSON serialization, where the `header`
properties are now nested inside a `LightClientHeader`.
The SSZ serialization remains compatible.
See https://github.com/ethereum/consensus-specs/pull/3190
and https://github.com/ethereum/beacon-APIs/pull/287
In a future fork, light client data will be extended with execution info
to support more use cases. To anticipate such an upgrade, introduce
`Forky` and `Forked` types, and ready the database schema.
Because the mapping of sync committee periods to fork versions is not
necessarily unique (fork schedule not in sync with period boundaries),
an additional column is added to `period` -> `LightClientUpdate` table.
* correctly report ignored contributions in metrics
* avoid counting subset contributions in vmon (bring in line with
attestation aggregates)
* avoid signature checks for subset attestations
A being a non-strict subset is a sufficient condition to ignore.
In order to avoid full replays when validating attestations hailing from
untaken forks, it's better to keep shufflings separate from `EpochRef`
and perform a lookahead on the shuffling when processing the block that
determines them.
This also helps performance in the case where REST clients are trying to
perform lookahead on attestation duties and decreases memory usage by
sharing shufflings between EpochRef instances of the same dependent
root.
Whether new blocks/attestations/etc are produced internally or received
via REST, their journey through the node is the same - to ensure that
they get the same treatment (logging, metrics, processing), this PR
moves the routing to a dedicated module and fixes several small
differences that existed before.
* `xxxValidator` -> `processMessageName` - the processor also was adding
messages to pools, so we want the name to reflect that action
* add missing "sent" metrics for some messages
* document ignore policy better - already-seen messages are not actaully
rebroadcast by libp2p
* skip redundant signature checks for internal validators consistently
The justified and finalized `Checkpoint` are frequently passed around
together. This introduces a new `FinalityCheckpoint` data structure that
combines them into one.
Due to the large usage of this structure in fork choice, also took this
opportunity to update fork choice tests to the latest v1.2.0-rc.1 spec.
Many additional tests enabled, some need more work, e.g. EL mock blocks.
Also implemented `discard_equivocations` which was skipped in #3661,
and improved code reuse across fork choice logic while at it.
Combines the LC data configuration options (serve / importMode), the
callbacks (finality / optimistic LC update) as well as the cache storing
light client data, into a new `LightClientDataStore` structure.
Also moves the structure into a light client specific file.
* document static vs dynamic range checking requirements
* add `vindices` iterator to iterate over valid validator indices in a
state
* clean up spec comments in general
* fixup
Co-authored-by: tersec <tersec@users.noreply.github.com>
Incorporates the latest changes to the light client sync protocol based
on Devconnect AMS feedback. Note that this breaks compatibility with the
previous prototype, due to changes to data structures and endpoints.
See https://github.com/ethereum/consensus-specs/pull/2802
Some upstream repos still need fixes, but this gets us close enough that
style hints can be enabled by default.
In general, "canonical" spellings are preferred even if they violate
nep-1 - this applies in particular to spec-related stuff like
`genesis_validators_root` which appears throughout the codebase.
Up til now, the block dag has been using `BlockRef`, a structure adapted
for a full DAG, to represent all of chain history. This is a correct and
simple design, but does not exploit the linearity of the chain once
parts of it finalize.
By pruning the in-memory `BlockRef` structure at finalization, we save,
at the time of writing, a cool ~250mb (or 25%:ish) chunk of memory
landing us at a steady state of ~750mb normal memory usage for a
validating node.
Above all though, we prevent memory usage from growing proportionally
with the length of the chain, something that would not be sustainable
over time - instead, the steady state memory usage is roughly
determined by the validator set size which grows much more slowly. With
these changes, the core should remain sustainable memory-wise post-merge
all the way to withdrawals (when the validator set is expected to grow).
In-memory indices are still used for the "hot" unfinalized portion of
the chain - this ensure that consensus performance remains unchanged.
What changes is that for historical access, we use a db-based linear
slot index which is cache-and-disk-friendly, keeping the cost for
accessing historical data at a similar level as before, achieving the
savings at no percievable cost to functionality or performance.
A nice collateral benefit is the almost-instant startup since we no
longer load any large indicies at dag init.
The cost of this functionality instead can be found in the complexity of
having to deal with two ways of traversing the chain - by `BlockRef` and
by slot.
* use `BlockId` instead of `BlockRef` where finalized / historical data
may be required
* simplify clearance pre-advancement
* remove dag.finalizedBlocks (~50:ish mb)
* remove `getBlockAtSlot` - use `getBlockIdAtSlot` instead
* `parent` and `atSlot` for `BlockId` now require a `ChainDAGRef`
instance, unlike `BlockRef` traversal
* prune `BlockRef` parents on finality (~200:ish mb)
* speed up ChainDAG init by not loading finalized history index
* mess up light client server error handling - this need revisiting :)
One more step on the journey to reduce `BlockRef` usage across the
codebase - this one gets rid of `StateData` whose job was to keep track
of which block was last assigned to a state - these duties have now been
taken over by `latest_block_root`, a fairly recent addition that
computes this block root from state data (at a small cost that should be
insignificant)
99% mechanical change.
* fewer deps on `BlockRef` traversal in anticipation of pruning
* allows identifying EpochRef:s by their shuffling as a first step of
* tighten error handling around missing blocks
using the zero hash for signalling "missing block" is fragile and easy
to miss - with checkpoint sync now, and pruning in the future, missing
blocks become "normal".
https://github.com/ethereum/consensus-specs/pull/2225 removed an ignore
rule that would filter out duplicate aggregates from gossip publishing -
however, this causes increased bandwidth and CPU usage as discussed in
https://github.com/ethereum/consensus-specs/issues/2183 - the intent is
to revert the removal and reinstate the rule.
This PR implements ignore filtering which cuts down on CPU usage (fewer
aggregates to validate) and bandwidth usage (less fanout of duplicates)
- as #2225 points out, this may lead to a small increase in IHAVE
messages.
* Harden handling of unviable forks
In our current handling of unviable forks, we allow peers to send us
blocks that come from a different fork - this is not necessarily an
error as it can happen naturally, but it does open up the client to a
case where the same unviable fork keeps getting requested - rather than
allowing this to happen, we'll now give these peers a small negative
score - if it keeps happening, we'll disconnect them.
* keep track of unviable forks in quarantine, to avoid filling it with
known junk
* collect peer scores in single module
* descore peers when they send unviable blocks during sync
* don't give score for duplicate blocks
* increase quarantine size to a level that allows finality to happen
under optimal conditions - this helps avoid downloading the same blocks
over and over in case of an unviable fork
* increase initial score for new peers to make room for one more failure
before disconnection
* log and score invalid/unviable blocks in requestmanager too
* avoid ChainDAG dependency in quarantine
* reject gossip blocks with unviable parent
* continue processing unviable sync blocks in order to build unviable
dag
* docs
* Update beacon_chain/consensus_object_pools/block_pools_types.nim
* add unviable queue test
* limit by-root requests to non-finalized blocks
Presently, we keep a mapping from block root to `BlockRef` in memory -
this has simplified reasoning about the dag, but is not sustainable with
the chain growing.
We can distinguish between two cases where by-root access is useful:
* unfinalized blocks - this is where the beacon chain is operating
generally, by validating incoming data as interesting for future fork
choice decisions - bounded by the length of the unfinalized period
* finalized blocks - historical access in the REST API etc - no bounds,
really
In this PR, we limit the by-root block index to the first use case:
finalized chain data can more efficiently be addressed by slot number.
Future work includes:
* limiting the `BlockRef` horizon in general - each instance is 40
bytes+overhead which adds up - this needs further refactoring to deal
with the tail vs state problem
* persisting the finalized slot-to-hash index - this one also keeps
growing unbounded (albeit slowly)
Anyway, this PR easily shaves ~128mb of memory usage at the time of
writing.
* No longer honor `BeaconBlocksByRoot` requests outside of the
non-finalized period - previously, Nimbus would generously return any
block through this libp2p request - per the spec, finalized blocks
should be fetched via `BeaconBlocksByRange` instead.
* return `Opt[BlockRef]` instead of `nil` when blocks can't be found -
this becomes a lot more common now and thus deserves more attention
* `dag.blocks` -> `dag.forkBlocks` - this index only carries unfinalized
blocks from now - `finalizedBlocks` covers the other `BlockRef`
instances
* in backfill, verify that the last backfilled block leads back to
genesis, or panic
* add backfill timings to log
* fix missing check that `BlockRef` block can be fetched with
`getForkedBlock` reliably
* shortcut doppelganger check when feature is not enabled
* in REST/JSON-RPC, fetch blocks without involving `BlockRef`
* fix dag.blocks ref
Time in the beacon chain is expressed relative to the genesis time -
this PR creates a `beacon_time` module that collects helpers and
utilities for dealing the time units - the new module does not deal with
actual wall time (that's remains in `beacon_clock`).
Collecting the time related stuff in one place makes it easier to find,
avoids some circular imports and allows more easily identifying the code
actually needs wall time to operate.
* move genesis-time-related functionality into `spec/beacon_time`
* avoid using `chronos.Duration` for time differences - it does not
support negative values (such as when something happens earlier than it
should)
* saturate conversions between `FAR_FUTURE_XXX`, so as to avoid
overflows
* fix delay reporting in validator client so it uses the expected
deadline of the slot, not "closest wall slot"
* simplify looping over the slots of an epoch
* `compute_start_slot_at_epoch` -> `start_slot`
* `compute_epoch_at_slot` -> `epoch`
A follow-up PR will (likely) introduce saturating arithmetic for the
time units - this is merely code moves, renames and fixing of small
bugs.
* Harden CommitteeIndex, SubnetId, SyncSubcommitteeIndex
Harden the use of `CommitteeIndex` et al to prevent future issues by
using a distinct type, then validating before use in several cases -
datatypes in spec are kept simple though so that invalid data still can
be read.
* fix invalid epoch used in REST
`/eth/v1/beacon/states/{state_id}/committees` committee length (could
return invalid data)
* normalize some variable names
* normalize committee index loops
* fix `RestAttesterDuty` to use `uint64` for `validator_committee_index`
* validate `CommitteeIndex` on ingress in REST API
* update rest rules with stricter parsing
* better REST serializers
* save lots of memory by not using `zip` ...at least a few bytes!
With checkpoint sync in particular, and state pruning in the future,
loading states or state-dependent data may fail. This PR adjusts the
code to allow this to be handled gracefully.
In particular, the new availability assumption is that states are always
available for the finalized checkpoint and newer, but may fail for
anything older.
The `tail` remains the point where state loading de-facto fails, meaning
that between the tail and the finalized checkpoint, we can still get
historical data (but code should be prepared to handle this as an
error).
However, to harden the code against long replays, several operations
which are assumed to work only with non-final data (such as gossip
verification and validator duties) now limit their search horizon to
post-finalized data.
* harden several state-dependent operations by logging an error instead
of introducing a panic when state loading fails
* `withState` -> `withUpdatedState` to differentiate from the other
`withState`
* `updateStateData` can now fail if no state is found in database - it
is also hardened against excessively long replays
* `getEpochRef` can now fail when replay fails
* reject blocks with invalid target root - they would be ignored
previously
* fix recursion bug in `isProposed`
Validator monitoring based on and mostly compatible with the
implementation in Lighthouse - tracks additional logs and metrics for
specified validators so as to stay on top on performance.
The implementation works more or less the following way:
* Validator pubkeys are singled out for monitoring - these can be
running on the node or not
* For every action that the validator takes, we record steps in the
process such as messages being seen on the network or published in the
API
* When the dust settles at the end of an epoch, we report the
information from one epoch before that, which coincides with the
balances being updated - this is a tradeoff between being correct
(waiting for finalization) and providing relevant information in a
timely manner)
* batch-verify sync messages for a small perf boost
Generally reuses the same structure as attestation and aggregate
verification
* normalize `signatures` and `signature_batch` to use the same pattern
of verification
* normalize parameter names, order etc for signature stuff in general
* avoid calling `blsSign` directly - instead, go through `signatures`
consistently
* move quarantine outside of chaindag
The quarantine has been part of the ChainDAG for the longest time, but
this design has a few issues:
* the function in which blocks are verified and added to the dag becomes
reentrant and therefore difficult to reason about - we're currently
using a stateful flag to work around it
* quarantined blocks bypass the processing queue leading to a processing
stampede
* the quarantine flow is unsuitable for orphaned attestations - these
should also should be quarantined eventually
Instead of processing the quarantine inside ChainDAG, this PR moves
re-queueing to `block_processor` which already is responsible for
dealing with follow-up work when a block is added to the dag
This sets the stage for keeping attestations in the quarantine as well.
Also:
* make `BlockError` `{.pure.}`
* avoid use of `ValidationResult` in block clearance (that's for gossip)
Renames and cleanups split out from the validator monitoring branch, so
as to reduce conflict area vs other PR:s
* add constants for expected message timing
* name validators after the messages they validate, mostly, to make
grepping easier
* unify field naming of EpochInfo across forks to make cross-fork code
easier
* `SyncCommitteeIndex` -> `SyncSubcommitteeIndex`
* `syncCommitteePeriod` -> `sync_committee_period` (spec spelling)
* tighten period comparisons
* fix assert when validating committee message with non-altair state in
REST api
There were still a few instances that used the expansion of `errReject`
instead of using the template itself. It seems that those cases were
forgotten as part of other cleanups in #2809. Done now for readability.
When sync committee message handling was introduced in #2830, the edge
case of the same validator being selected multiple times as part of a
sync subcommittee was not covered. Not handling that edge case makes
sync contributions have a lower-than-expected participation rate as each
sync validator is only counted up through once per subcommittee.
This patch ensures that this edge case is properly covered.
The P2P spec defines how certain error classes should be handled through
either IGNORE or REJECT verdicts. For sync committee message, the spec
defines that only the first message from each validator per subcommittee
and slot shall be accepted, the rest is ignored. However, current code
rejects those messages instead of ignoring them. Fixed to match spec.
* Placing callbacks into strategic places.
* Initial events call implementation.
* Post rebase fixes.
* Change addSyncContribution() implementation.
* Add `attestation-sent` event.
Remove gcsafe, raises from callbacks implementations.
Move `attestation-received` fire at the end of attestation processing.
* Address review comments.
* cleanups
* use ForkedTrustedSignedBeaconBlock.ionit where appropriate
* move `is_aggregator` to `spec/`
* use `errReject` in a few more places
* update enr fork id when time is auspicious
* use network broadcast functions
* Return Ignore for aggregate signature validation timeouts
...consistently between aggregates and attestations.
* clean up some more reject/ignore rules
* shorten texts a bit
* errReject->checkedReject, use err helpers throughout
* get rid of quarantine in exitpool as well
* reorganize ssz dependencies
This PR continues the work in
https://github.com/status-im/nimbus-eth2/pull/2646,
https://github.com/status-im/nimbus-eth2/pull/2779 as well as past
issues with serialization and type, to disentangle SSZ from eth2 and at
the same time simplify imports and exports with a structured approach.
The principal idea here is that when a library wants to introduce SSZ
support, they do so via 3 files:
* `ssz_codecs` which imports and reexports `codecs` - this covers the
basic byte conversions and ensures no overloads get lost
* `xxx_merkleization` imports and exports `merkleization` to specialize
and get access to `hash_tree_root` and friends
* `xxx_ssz_serialization` imports and exports `ssz_serialization` to
specialize ssz for a specific library
Those that need to interact with SSZ always import the `xxx_` versions
of the modules and never `ssz` itself so as to keep imports simple and
safe.
This is similar to how the REST / JSON-RPC serializers are structured in
that someone wanting to serialize spec types to REST-JSON will import
`eth2_rest_serialization` and nothing else.
* split up ssz into a core library that is independendent of eth2 types
* rename `bytes_reader` to `codec` to highlight that it contains coding
and decoding of bytes and native ssz types
* remove tricky List init overload that causes compile issues
* get rid of top-level ssz import
* reenable merkleization tests
* move some "standard" json serializers to spec
* remove `ValidatorIndex` serialization for now
* remove test_ssz_merkleization
* add tests for over/underlong byte sequences
* fix broken seq[byte] test - seq[byte] is not an SSZ type
There are a few things this PR doesn't solve:
* like #2646 this PR is weak on how to handle root and other
dontSerialize fields that "sometimes" should be computed - the same
problem appears in REST / JSON-RPC etc
* Fix a build problem on macOS
* Another way to fix the macOS builds
Co-authored-by: Zahary Karadjov <zahary@gmail.com>
The spec imports are a mess to work with, so this branch cleans them up
a bit to ensure that we avoid generic sandwitches and that importing
stuff generally becomes easier.
* reexport crypto/digest/presets because these are part of the public
symbol set of the rest of the spec types
* don't export `merge` types from `base` - this causes circular deps
* fix circular deps in `ssz/spec_types` - this is the first step in
disentangling ssz from spec
* be explicit about phase0 vs altair - longer term, `altair` will become
the "natural" type set, then merge and so on, so no point in giving
`phase0` special preferential treatment
Simpler module name for stuff that covers forks
* check that runtime config matches database state
* also include some assorted altair cleanups
* use "standard" genesis fork in local testnet to work around missing
runtime config support
* Implement split preset/config support
This is the initial bulk refactor to introduce runtime config values in
a number of places, somewhat replacing the existing mechanism of loading
network metadata.
It still needs more work, this is the initial refactor that introduces
runtime configuration in some of the places that need it.
The PR changes the way presets and constants work, to match the spec. In
particular, a "preset" now refers to the compile-time configuration
while a "cfg" or "RuntimeConfig" is the dynamic part.
A single binary can support either mainnet or minimal, but not both.
Support for other presets has been removed completely (can be readded,
in case there's need).
There's a number of outstanding tasks:
* `SECONDS_PER_SLOT` still needs fixing
* loading custom runtime configs needs redoing
* checking constants against YAML file
* yeerongpilly support
`build/nimbus_beacon_node --network=yeerongpilly --discv5:no --log-level=DEBUG`
* load fork epoch from config
* fix fork digest sent in status
* nicer error string for request failures
* fix tools
* one more
* fixup
* fixup
* fixup
* use "standard" network definition folder in local testnet
Files are loaded from their standard locations, including genesis etc,
to conform to the format used in the `eth2-networks` repo.
* fix launch scripts, allow unknown config values
* fix base config of rest test
* cleanups
* bundle mainnet config using common loader
* fix spec links and names
* only include supported preset in binary
* drop yeerongpilly, add altair-devnet-0, support boot_enr.yaml
* use ForkedHashedBeaconState in StateData
* fix FAR_FUTURE_EPOCH -> slot overflow; almost always use assign()
* avoid stack allocation in maybeUpgradeStateToAltair()
* create and use dispatch functions for check_attester_slashing(), check_proposer_slashing(), and check_voluntary_exit()
* use getStateRoot() instead of various state.data.hbsPhase0.root
* remove withStateVars.hashedState(), which doesn't work as a design anymore
* introduce spec/datatypes/altair into beacon_chain_db
* fix inefficient codegen for getStateField(largeStateField)
* state_transition_slots() doesn't either need/use blocks or runtime presets
* combine process_slots(HBS)/state_transition_slots(HBS) which differ only in last-slot htr optimization
* getStateField(StateData, ...) was replaced by getStateField(ForkedHashedBeaconState, ...)
* fix rollback
* switch some state_transition(), process_slots, makeTestBlocks(), etc to use ForkedHashedBeaconState
* remove state_transition(phase0.HashedBeaconState)
* remove process_slots(phase0.HashedBeaconState)
* remove state_transition_block(phase0.HashedBeaconState)
* remove unused callWithBS(); separate case expression from if statement
* switch back from nested-ref-object construction to (ref Foo)(Bar())
* write uncompressed validator keys to database
Loading 150k+ validator keys on startup in compressed format takes a lot
of time - better store them in uncompressed format which makes behaviour
just after startup faster / more predictable.
* refactor cached validator key access
* fix isomorphic cast to work with non-var instances
* remove cooked pubkey cache - directly use database cache in chaindag
as well (one less cache to keep in sync)
* bump blscurve, introduce loadValid for known-to-be-valid keys
Instead of keeping a validator key list per EpochRef, this PR introduces
a single shared validator key list in ChainDAG, and cleans up some other
ChainDAG and key-related issues.
The PR does not introduce the validator key list in the state transition
- this is because we batch-check all signatures before entering the spec
code, thus the spec code never hits the cache.
A future refactor should _probably_ remove the threadvar altogether.
There's a few other small fixes in here that make the flow easier to
read:
* fix `var ChainDAGRef` -> `ChainDAGRef`
* fix `var QuarantineRef` -> `QuarantineRef`
* consistent `dag` variable name
* avoid using threadvar pubkey cache in most cases
* better error messages in batch signature checking
Currently, we have a bit of a convoluted flow where when sending
attestations, we start broadcasting them over gossip then pass them to
the attestation validation to include them in the local attestation pool
- it should be the other way around: we should be checking attestations
_before_ gossipping them - this serves as an additional safety net to
ensure that we don't publish junk - this becomes more important when
publishing attestations from the API.
Also, the REST API was performing its own validation meaning
attestations coming from REST would be validated twice - finally, the
JSON RPC wasn't pre-validating and would happily broadcast invalid
attestations.
* Unified attestation production pipeline with the same flow for gossip,
locally and API-produced attestations: all are now validated and entered
into the pool, then broadcast/republished
* Refactor subnet handling with specific SubnetId alias, streamlining
where subnets are computed, avoiding the need to pass around the number
of active validators
* Move some of the subnet handling code to eth2_network
* Use BitArray throughout for subnet handling
With the introduction of batching and lazy attestation aggregation, it
no longer makes sense to enqueue attestations between the signature
check and adding them to the attestation pool - this only takes up
valuable CPU without any real benefit.
* add successfully validated attestations to attestion pool directly
* avoid copying participant list around for single-vote attestations,
pass single validator index instead
* release decompressed gossip memory earlier, specially during async
message validation
* use cooked signatures in a few more places to avoid reloads and errors
* remove some Defect-raising versions of signature-loading
* release decompressed data memory before validating message
* avoid creating indexed attestation just to check signatures - above
all, don't create it when not checking signatures ;)
* avoid pointer op when adding attestation to pool
* better iterator for yielding attestations
* add metric / log for attestation packing time
This is a revamp of the attestation pool that cleans up several aspects
of attestation processing as the network grows larger and block space
becomes more precious.
The aim is to better exploit the divide between attestation subnets and
aggregations by keeping the two kinds separate until it's time to either
produce a block or aggregate. This means we're no longer eagerly
combining single-vote attestations, but rather wait until the last
moment, and then try to add singles to all aggregates, including those
coming from the network.
Importantly, the branch improves on poor aggregate quality and poor
attestation packing in cases where block space is running out.
A basic greed scoring mechanism is used to select attestations for
blocks - attestations are added based on how much many new votes they
bring to the table.
* Collect single-vote attestations separately and store these until it's
time to make aggregates
* Create aggregates based on single-vote attestations
* Select _best_ aggregate rather than _first_ aggregate when on
aggregation duty
* Top up all aggregates with singles when it's time make the attestation
cut, thus improving the chances of grabbing the best aggregates out
there
* Improve aggregation test coverage
* Improve bitseq operations
* Simplify aggregate signature creation
* Make attestation cache temporary instead of storing it in attestation
pool - most of the time, blocks are not being produced, no need to keep
the data around
* Remove redundant aggregate storage that was used only for RPC
* Use tables to avoid some linear seeks when looking up attestation data
* Fix long cleanup on large slot jumps
* Avoid some pointers
* Speed up iterating all attestations for a slot (fixes#2490)
* only deserialize attestation and aggregation gossiped signatures once
* re-indent some aggregate checks into block scope
* spelling
* remove debugging assertion
* put part of gossip validation back into block context
* attestation pool test signature loading isn't so unsafe, and exportRaw isn't free
* remove more development doAsserts; don't exportRaw in loops
* batch attestations
* Fixes (but now need to investigate the chronos 0 .. 4095 crash similar to https://github.com/status-im/nimbus-eth2/issues/1518
* Try to remove the processing loop to no avail :/
* batch aggregates
* use resultsBuffer size for triggering deadline schedule
* pass attestation pool tests
* Introduce async gossip validators. May fix the 4096 bug (reentrancy issue?) (similar to sync unknown blocks #1518)
* Put logging at debug level, add speed info
* remove unnecessary batch info when it is known to be one
* downgrade some logs to trace level
* better comments [skip ci]
* Address most review comments
* only use ref for async proc
* fix exceptions in eth2_network
* update async exceptions in gossip_validation
* eth2_network 2nd pass
* change to sleepAsync
* Update beacon_chain/gossip_processing/batch_validation.nim
Co-authored-by: Jacek Sieka <jacek@status.im>
Co-authored-by: Jacek Sieka <jacek@status.im>
Etan Kissling
Jacek Sieka
tersec
henridf
zah
Miran
Zahary Karadjov
Eugene Kabanov
Mamy André-Ratsimbazafy