Getting not in memory pool stake wins. v0.22.14 and v0.22.15 - possible code regression.

[the-real-vortex-v]

Describe the bug

Failed stakes not auto aborting like they did in the last few versions. I think a code regression has taken place.
Getting stake wins that end up as "not in memory pool" like the below after updating to v0.22.14:

Status: 0/unconfirmed, not in memory pool
Date: 4/10/2025 11:16
From: unknown

I just got another today.

Status: 0/unconfirmed, not in memory pool
Date: 4/10/2025 19:09
From: unknown

That's 3 in total since the 0.22.14 node update.

To Reproduce

Run the node on windows 10.
Wait for stake to happen. So far it's about ~15% of stakes. I've had 2 in the last 2 days.
3 good stakes and 2 not in memory pool.
I abandoned the below stake (4/9/2025) in red (in the screen shot).

Restarting the node didn't clear it.
Expected behavior
Abandoned stakes should clear automatically. This seems like a regression in the code that was fixed several versions ago.

Screenshots

Desktop (please complete the following information):
Windows 10

[the-real-vortex-v]

Just a quick update. I'm getting 1 to 3 of these a day at the moment. I'm not stopping my node and trying run any kind of wallet scan right now. I think there's around 10 failed stakes stuck like this at the moment.

[andyoknen]

I found a problem with one important table (SocialRegistry), which usually occurs after an abnormal shutdown of the node. I've been testing this version for the last few days. This problem causes undesirable behavior in staking - the node may include unwanted transactions in the blocks being created (if the table is corrupted), which other nodes will not accept. I think this problem is related to some of the "bad" stakes.

[the-real-vortex-v]

Here is an update on the issue for me:

Can I fix this or is this start a new wallet time?

[andyoknen]

I am currently considering the possibility of automatically canceling orphan stacking transactions. I will also return to studying the problem after the release of 0.22.16, which will fix the problem with the SocialRegistry table.