Martin HS

I'm sceptical to the idea that we should do flush from the outside. IMO the basic approach should be - Always immediately flush leafs (and obviously calc the committment for...

For lack of a better place to post this, I'll just post it here. It's not directly related to this PR as is right now, but probably a future version...

RIght, no that's true, it's not settabie, not via config nor cli. It's calculated on the fly,

I think a valid solution would be, not to set the max database handles explicitly, but rather to override the max file limit. So instead of geth checking the max...

Just curious, instead of padding, we could just return a smaller slice than requested. How come it was decided to zero-pad instead?

No idea why, but appveyor 32-bit build says ``` t=2022-08-31T06:34:34+0000 lvl=trce msg="Promoted queued transactions" count=0 t=2022-08-31T06:34:34+0000 lvl=info msg="Transaction pool stopped" runtime: out of memory: cannot allocate 4194304-byte block (1713668096 in...

@jwasinger would you mind rebasing this?

Testing this on two bootnodes which have a very hard time syncing ``` ansible-playbook playbook.yaml -t geth -l bootnode-azure-westus-001,bootnode-azure-koreasouth-001 -e "geth_image=holiman/geth-experimental:latest" -e "geth_datadir_wipe=partial" -e '{"geth_args_custom":["--backingdb=pebble"]}' ``` and for comparison: ```...

Metrics that do not appear to work: `geth.eth/db/chaindata/disk/read.meter` `geth.eth/db/chaindata/compact/writedelay/counter.meter` Edi - here's why the disk read meter is always zerot: ```golang if db.diskReadMeter != nil { db.diskReadMeter.Mark(0) // pebble doesn't...

It's still very early, but it looks like on our 'weak' azure nodes, pebble is performing a lot better, since it's not being killed-by-compaction: ![Screenshot 2022-08-29 at 16-07-01 Dual Geth...