[bug]: All channels inactive after restart: Unable to retrieve mempool txs #7778
Description
Background
After restarting my lnd node, it seems to be unable to pull data from bitcoind.
I have tried with two different bitcoin nodes, both of which have been synced and are able to service other lnd nodes.
Your environment
lncli version 0.16.3-beta commit=v0.16.3-beta
6.3.0-1-armmp-lpae #1 SMP Debian 6.3.7-1 (2023-06-12) armv7l GNU/Linux
Steps to reproduce
Restart lnd
Expected behaviour
The server syncs chain state and then connects to channels
Actual behaviour
2023-06-19 23:59:05.154 [INF] LTND: Waiting for chain backend to finish sync, start_height=795134
2023-06-19 23:59:56.718 [INF] LNWL: Started rescan from block 000000000000000000003c4f88f639857342b68d8aa10a4fa3ae81
0ed2fcb749 (height 795068) for 183 addresses
2023-06-20 00:00:35.083 [ERR] LNWL: Unable to complete chain rescan: error reading json reply: unexpected EOF
2023-06-20 00:03:24.996 [WRN] LNWL: unable to fetch transaction c6caca7a713fd44a00186a294f9b654499b2324c1201163fe678
11a0f3956a25 for mempool: -5: No such mempool or blockchain transaction. Use gettransaction for wallet transactions.
....
more unable to fetch transactions (most of which have been RBF)
...
2023-06-20 01:01:42.979 [INF] LNWL: Started polling mempool to cache new transactions
2023-06-20 01:03:02.600 [ERR] LNWL: Unable to retrieve mempool txs: error reading json reply: unexpected EOF
2023-06-20 01:03:37.511 [ERR] LNWL: Unable to retrieve mempool txs: error reading json reply: unexpected EOF
With the unexpected EOF just happening forever.
I've had the lnd node running for about 24 hours since the restart and it is still just showing EOF approximately every minute or two.