It is common for a validator operator to connect their validator
only to other nodes under their control, using clustering. Relaying of
untrusted validations and proposals must be unreliable to prevent
denial of service attacks. But currently, they are unreliable even
within a cluster.
With this change, a cluster member's decisison to relay (or originate)
a validation or proposal is honored by other cluster members. This
ensures that validators in a cluster will get reliable relaying to
hubs outside the cluster, even if other members of the cluster do not
have that validator on their UNL.
* Changes to Travis seem to be making these builds slower, and are
resulting in spurious timeouts. Remove the builds that fail more
often than not until we complete our migration to another build
system.
* Use CMake for unity builds, instead of the slow build.
* Can be exercised from the command line with json2
* Rewrite Env::do_rpc to call the same code as
rpc from the command line. This puts rpc
handling logic in one place.
* If any of the destructor, copy assignment or copy constructor
are user-declared, both copy members should be user-declared,
otherwise the compiler-generation of them is deprecated.
* If any of the destructor, copy assignment or copy constructor
are user-declared, both copy members should be user-declared,
otherwise the compiler-generation of them is deprecated.
* Remove composite helper functions
* Add set difference and Bitset/uint256 operators
* Convert tests to use new feature bitset set difference operator
In order to automatically run unit tests with newly created
amendments, prefer to start with jtx::supported_features() and
then subtract unwanted features.
These changes identified a few bugs that were hiding in
amendments. One of those bugs, in FlowCross, is not yet fixed.
By uncommenting the test in CrossingLimits_test.cpp you can see
failures relating to that bug. Since FlowCross is not yet
enabled on the network we can fix the bug at our convenience.
Both Tickets and SHAMapV2 have been around for a while and don't
look like they will be enabled on the network soon. So they are
removed from the supportedAmendments list. This prevents Env
from automatically testing with Tickets or SHAMapV2 enabled,
although testing with those features can still be explicitly
specified.
Drive-by cleanups:
o supportedAmendments() returns a const reference rather than
a fresh vector on each call.
o supportedAmendments() implementation moved from Amendments.cpp
to Feature.cpp. Amendments.cpp deleted.
o supportedAmendments() declared in Feature.h. All other
declarations deleted.
o preEnabledAmendments() removed, since it was empty and only
used in one place. It will be easy to re-add when it is needed.
o jtx::all_features_except() renamed to
jtx::supported_features_except(), which is more descriptive.
o jtx::all_amendments() renamed to jxt::supported_amendments()
o jtx::with_features() renamed to with_only_features()
o Env_test.cpp adjusted since featureTickets is no longer
automatically enabled for unit tests.
- Separate `Scheduler` from `BasicNetwork`.
- Add an event/collector framework for monitoring invariants and calculating statistics.
- Allow distinct network and trust connections between Peers.
- Add a simple routing strategy to support broadcasting arbitrary messages.
- Add a common directed graph (`Digraph`) class for representing network and trust topologies.
- Add a `PeerGroup` class for simpler specification of the trust and network topologies.
- Add a `LedgerOracle` class to ensure distinct ledger histories and simplify branch checking.
- Add a `Submitter` to send transactions in at fixed or random intervals to fixed or random peers.
Co-authored-by: Joseph McGee
We need to ensure that pointers and/or references to existing
elements will not be invalidated during the course of element
insertion and removal.
Containers like std::vector do not offer this guarantee, and
cannot be used as the underlying container for the stack.
By choosing to explicitly specify std::deque as the underlying
cotnainer, we avoid:
- the unlikely possibility of the C++ standards committee
changing the default template parameter to a container;
- the more likely possibility of an accidental change by
a programmer, without fully considering the consequences.
* Stores recent history of "good" ledgers. Uses the maximum as the
expected ledger size. When a large value drops off, use a 90%
backoff to go down to to the new maximum.
* If consensus is unhealthy, wipe the history in addition to the current
clamping.
* Include .md doc files in xcode and VS projects
In support of dynamic validator list, this changeset:
1. Adds a new `validator_list_expires` field to `server_info` that
indicates when the current validator list will become stale.
2. Adds a new admin only `validator_lists` RPC that returns the
current list of known validators and the most recent published validator
lists.
3. Adds a new admin only `validator_sites` RPC that returns the list of
configured validator publisher sites and when they were most recently
queried.
Manifests of validators newly added to a published validator list are
not reliably propagated to network nodes.
This solves the problem by allowing a published validator list to
include the manifest.
RIPD-1559
Switches the default behavior of Consensus to use roundCloseTime instead of
effCloseTime. effCloseTime is still used when accepting the consensus ledger to
ensure the consensus close time comes after the parent ledger close time. This
change eliminates an edge case in which peers could reach agreement on the close
time, but end up generating ledgers with different close times.
Introduce "fix1523" which corrects a minor technical flaw with
the original implementation of the escrow feature.
When creating an escrow, the entry would only be tracked in the
owner directory of the sender; as a result, an escrow recipient
would not be able to detect incoming escrows without monitoring
the ledger in real-time for transactions of interest or without
the sender communicating this information out of band.
With the fix in place, escrows where the recipient differs from
the sender will be listed in the recipient's owner directory as
well.
* If the transaction can't be queued, recover to the open ledger once,
and drop it on the next attempt.
* New result codes for transactions that can not queue.
* Add minimum queue size.
* Remove the obsolete and incorrect SF_RETRY flag.
* fix#2215
