Rescuing a zcashd node that won't catch up

Hi, I have a zcashd that won’t reindex/rescan fully and catch up. Any tips or tricks that I might try to recover the health of the node? Can I delete Viewing Keys that I don’t need? I have 2 notes on this kanban board. I can pay a small bounty for markdown guides in this repo. If such documentation is already available, where is it? What are the best places to find such guides, tips and tricks?

I’m interested in this topic of a body of guides, tips and tricks in general. But, I’ll include the specific problem I’m having here in case someone wants to take a stab at it in the comments.

zcashd: wallet/wallet.cpp:2945: void CWallet::DecrementNoteWitnesses(const Consensus::Params &, const CBlockIndex *): Assertion `pindex->pprev->hashFinalOrchardRoot == orchardWallet.GetLatestAnchor()' failed.

I found this issue: Crash in DecrementNoteWitnesses after a reorg · Issue #5960 · zcash/zcash · GitHub so I pasted some of the details there.

See also some discord chatter: Discord

And an earlier issue that led me to try reindex: Zcashd help

How fast is your CPU? Are you using an SSD?

Standard cloud hardware. I could look it up. Interestingly going from 2 vCPU cores to 4 doesnt seem to help, the process seems to peg 1 core at 100%. The drive is also pretty default but i noticed it was SCSI the other day. I could look up the specs but the with 4 cores, 16 gigs of memory and 200gb SCSI, i want to say that the server costs around $200/month.

FWIW, the 2 vcpu spec might be less than what the full node needs now to crunch through mainnet chains. Additionally the RAM requirement is around 8+GB now. I think your error is mentioned on githhub but might also be a result of resource constrains on a sync. Not sure if this info helps.

Adding vCPU cores doesnt seem to change anything. It appears that there is only one process at work that is not parallelized over extra cores. 2 may be better than 1 so the extra core can do regular OS stuff. But going from 2 to 4 doesnt help anything.