fromNetwork() is async — it posts handlers to the io_service and
returns immediately. The original sendThread() loop fires all queued
events as concurrent HTTP connections at once. Under sustained load
with a slow/failing endpoint, connections accumulate (each held up to
30s by RPC_NOTIFY timeout), exhausting file descriptors and breaking
all network I/O for the entire process.
Fix: use a local io_service per batch with .run() to block until the
batch completes (same pattern as rpcClient() in RPCCall.cpp). This
bounds concurrent connections to maxInFlight (32) per subscriber while
still allowing parallel delivery.
Also add a queue cap (maxQueueSize = 16384, ~80-160MB) so a hopelessly
behind endpoint doesn't grow the deque indefinitely. Consumers detect
gaps via the existing seq field.
Ref: XRPLF/rippled#6341
Fixes the use of high and low in variable names, as these are determined by ripple::keylet::line processing.
Co-authored-by: RichardAH <richard.holland@starstone.co.nz>
std::endl triggers flush() which calls sync() on the shared log buffer.
Multiple threads racing in sync() cause str()/str("") operations to
corrupt buffer state, leading to crashes and double frees.
Added mutex to serialize access to suite.log, preventing concurrent
sync() calls on the same buffer.
* Match unit tests on start of test name (#4634)
* For example, without this change, to run the TxQ tests, must specify
`--unittest=TxQ1,TxQ2` on the command line. With this change, can use
`--unittest=TxQ`, and both will be run.
* An exact match will prevent any further partial matching.
* This could have some side effects for different tests with a common
name beginning. For example, NFToken, NFTokenBurn, NFTokenDir. This
might be useful. If not, the shorter-named test(s) can be renamed. For
example, NFToken to NFTokens.
* Split the NFToken, NFTokenBurn, and Offer test classes. Potentially speeds
up parallel tests by a factor of 5.
* SetHook_test, SetHookTSH_test, XahauGenesis_test
---------
Co-authored-by: Ed Hennis <ed@ripple.com>
