Same failing route being attempted over and over again for a given payment

[microlancer]

My node is stuck making the same failing route request over and over for a given invoice payment.

| HTLC_STATE                          | ATTEMPT_TIME | RESOLVE_TIME | RECEIVER_AMT | FEE  | TIMELOCK | CHAN_OUT           | ROUTE                                                                                             |
+-------------------------------------+--------------+--------------+--------------+------+----------+--------------------+---------------------------------------------------------------------------------------------------+
| TEMPORARY_CHANNEL_FAILURE @ 2nd hop |        0.983 |        7.096 | 10000        | 3.51 |   823206 | 864405255643201536 | PLEB.DEGEN->PiLN-Chiba\U0001f333Magma Ready\U0001f30b->LNB\u03dfG [Edge-4]->LNB\u03dfG [Hub-2]->Altcoins.bg->microlancer.io |
| TEMPORARY_CHANNEL_FAILURE @ 2nd hop |        8.129 |       14.052 | 10000        | 3.51 |   823206 | 864405255643201536 | PLEB.DEGEN->PiLN-Chiba\U0001f333Magma Ready\U0001f30b->LNB\u03dfG [Edge-4]->LNB\u03dfG [Hub-2]->Altcoins.bg->microlancer.io |
| TEMPORARY_CHANNEL_FAILURE @ 2nd hop |       15.088 |       22.586 | 10000        | 3.51 |   823206 | 864405255643201536 | PLEB.DEGEN->PiLN-Chiba\U0001f333Magma Ready\U0001f30b->LNB\u03dfG [Edge-4]->LNB\u03dfG [Hub-2]->Altcoins.bg->microlancer.io |
| TEMPORARY_CHANNEL_FAILURE @ 2nd hop |       23.597 |       31.390 | 10000        | 3.51 |   823206 | 864405255643201536 | PLEB.DEGEN->PiLN-Chiba\U0001f333Magma Ready\U0001f30b->LNB\u03dfG [Edge-4]->LNB\u03dfG [Hub-2]->Altcoins.bg->microlancer.io |
+-------------------------------------+--------------+--------------+--------------+------+----------+--------------------+---------------------------------------------------------------------------------------------------+
Amount + fee:   0 + 0 sat
Payment hash:   05e44ee6607e8340824e3f8b7ad5d700eb445776d515e4dd0b3dfb676939ca40
Payment status: FAILED, reason: FAILURE_REASON_TIMEOUT
[lncli] FAILED

prod:/# lncli getmccfg
{
    "config":  {
        "half_life_seconds":  "3600",
        "hop_probability":  0.6,
        "weight":  0.5,
        "maximum_payment_results":  500000,
        "minimum_failure_relax_interval":  "3600",
        "model":  "APRIORI",
        "apriori":  {
            "half_life_seconds":  "3600",
            "hop_probability":  0.6,
            "weight":  0.5,
            "capacity_fraction":  0.9999
        }
    }
}

Is there any way to make LND skip known failing paths so that it doesn't keep repeating the same failure?

I checked lncli querymc and searched for the particular pair that is failing here that is PiLN-Chiba to LNBIG-Edge-4 and based on the node IDs, there is no entry indicating that the pair's to/from are failing. That might actually be a clue, that somehow LND isn't successfully writing the failure to the database for mission control. Maybe it's some really rare edge case.

I also checked to see if I hit the max 500,000 payment results but it doesn't seem like it at all based on this:

# lncli querymc | grep from | wc -l
1364

This keeps happening to me for a variety of payments, and very often. I'll keep trying to dig into the logs but it's always difficult to figure out what part of the logs is relevant because debug output is so verbose.

I think not only is it wasting resources on my end trying the same failing path over and over, but also for others on the network. Am I missing something obvious?

[microlancer]

Also, I'm not using bimodal but would be curious if anything in #7520 could be affecting this.

[yyforyongyu]

Which lnd version are you using?

[microlancer]

I am using LND 0.17.3-beta and bitcoind 26.0