[bug]: maximum `max_events` documentation for `ForwardingHistory` not actually implemented

[ZZiigguurraatt]

Pre-Submission Checklist

• I have searched the existing issues and believe this is a new bug.
• I am not asking a question about how to use lnd, but reporting a bug (otherwise open a discussion).

LND Version

lncli version 0.20.0-beta commit=lightning-terminal-v0.16.0-alpha

LND Configuration

            - --disableui
            - --macaroonpath=/home/litd/.lit/regtest/lit.macaroon
            - --lnd.bitcoind.zmqpubrawtx=tcp://bitcoind:28335
            - --lnd.bitcoind.zmqpubrawblock=tcp://bitcoind:28334
            - --lnd.bitcoin.regtest
            - --lnd-mode=integrated
            - --lnd.allow-circular-route
            - --pool-mode=disable
            - --loop-mode=disable
            - --autopilot.disable
            - --lnd.debuglevel=debug
            - --lnd.trickledelay=50
            - --lnd.gossip.sub-batch-delay=5ms
            - --lnd.caches.rpc-graph-cache-duration=100ms
            - --lnd.default-remote-max-htlcs=483
            - --lnd.dust-threshold=5000000
            - --lnd.protocol.option-scid-alias
            - --lnd.protocol.zero-conf
            - --lnd.protocol.simple-taproot-chans
            - --lnd.protocol.simple-taproot-overlay-chans
            - --lnd.protocol.wumbo-channels
            - --lnd.accept-keysend
            - --lnd.protocol.custom-message=17
            - --lnd.wallet-unlock-password-file=/tmp/wallet-password
            - --lnd.wallet-unlock-allow-create
            - --taproot-assets.allow-public-uni-proof-courier
            - --taproot-assets.universe.public-access=rw
            - --taproot-assets.universe.sync-all-assets
            - --taproot-assets.allow-public-stats
            - --taproot-assets.universerpccourier.skipinitdelay
            - --taproot-assets.universerpccourier.backoffresetwait=1s
            - --taproot-assets.universerpccourier.numtries=5
            - --taproot-assets.universerpccourier.initialbackoff=300ms
            - --taproot-assets.universerpccourier.maxbackoff=600ms
            - --taproot-assets.experimental.rfq.acceptpricedeviationppm=100000
            - --taproot-assets.experimental.rfq.priceoracleaddress=use_mock_price_oracle_service_promise_to_not_use_on_mainnet
            - --taproot-assets.experimental.rfq.mockoracleassetsperbtc=10000000000

Backend Version

bitcoind 29.1

Backend Configuration

            - -server=1
            - -regtest=1
            - -zmqpubrawblock=tcp://0.0.0.0:28334
            - -zmqpubrawtx=tcp://0.0.0.0:28335
            - -zmqpubhashblock=tcp://0.0.0.0:28336
            - -zmqpubhashblockhwm=100000
            - -zmqpubhashtxhwm=100000
            - -zmqpubrawblockhwm=100000
            - -zmqpubrawtxhwm=100000
            - -txindex=1
            - -dnsseed=0
            - -upnp=0
            - -rpcbind=0.0.0.0
            - -rpcallowip=0.0.0.0/0
            - -rpcthreads=40
            - -rest
            - -listen=1
            - -listenonion=0
            - -fallbackfee=0.0002
            - -blockfilterindex=1
            - -peerblockfilters=1

OS/Distribution

Ubuntu 22.04.5 LTS

Bug Details & Steps to Reproduce

At

• lnd/rpcserver.go

  Lines 8193 to 8198 in 7bbf26c

  	// A list of forwarding events are returned. The size of each forwarding event
  	// is 40 bytes, and the max message size able to be returned in gRPC is 4 MiB.
  	// In order to safely stay under this max limit, we'll return 50k events per
  	// response. Each response has the index offset of the last entry. The index
  	// offset can be provided to the request to allow the caller to skip a series
  	// of records.

• lnd/lnrpc/lightning.proto

  Lines 471 to 478 in 7bbf26c

  	A list of forwarding events are returned. The size of each forwarding event
  	is 40 bytes, and the max message size able to be returned in gRPC is 4 MiB.
  	As a result each message can only contain 50k entries. Each response has
  	the index offset of the last entry. The index offset can be provided to the
  	request to allow the caller to skip a series of records.
  	*/
  	rpc ForwardingHistory (ForwardingHistoryRequest)
  	returns (ForwardingHistoryResponse);

• lnd/cmd/commands/cmd_payments.go

  Lines 1545 to 1546 in 7bbf26c

  	The max number of events returned is 50k. The default number is 100,
  	callers can use the --max_events param to modify this value.

it says the maximum number of records returned is 50,000.

I can't find anywhere in the code that this is actually enforced and when testing, it doesn't seem to be because I get

litd@james-886dd699c-4vb2q:/$ lncli fwdinghistory --max_events 999999999 --start_time -999d|tail -2
    "last_offset_index": 277736
}
litd@james-886dd699c-4vb2q:/$ 

Expected Behavior

max_events should be limited as the documentation says. If the user sets too large of a value, this is going to cause problems like is mentioned because the max size of a gRPC message, in which case we will get problems like #9709 (comment).

Debug Information

No response

Environment

litd in kubernetes

[ZZiigguurraatt]

Note, we should fix this issue after we fix #9882 , but in the same release.

[fresh3nough]

I've consolidated PR #10503 and PR #10512 into a single solution in #10512.

The consolidated approach combines the best of both PRs:

• Returns an error when the limit is exceeded (better API feedback)
• Uses the existing channeldb.MaxResponseEvents constant (avoids duplication)
• Includes comprehensive unit tests

This should fully resolve this issue.