bgelb

I guess a fairly obvious question is whether there is some kind of corruption in my db. Would be good to see if this can reproduced on another erigon instance...

My random sampling seems to have converged to a contiguous hole from block 15140000 to 15141999... exactly 2000 blocks. Does this ring any bells w.r.t. the internal structures? I am...

Update, 8 hours later, the contiguous "bad" range spans now from block 15140000 to 15143999 (+2000 blocks added). Note this is still statistical sampling... so its possible the numbers above...

Log after restart: ``` Jul 28 20:31:43 slc04 ERIGON[383187]: [INFO] [07-28|20:31:43.137] Starting metrics server addr=http://0.0.0.0:6060/debug/metrics/prometheus Jul 28 20:31:43 slc04 ERIGON[383187]: [INFO] [07-28|20:31:43.137] Build info git_branch=v2022.07.03-bgelb git_tag= git_commit=2f8f539daac12bad029963157d03495e81c58bb7 Jul 28 20:31:43...

Now seems to be 1000 blocks hole from 15144000 to 15145000...

Please advise what else you'd like to see in erigon log...

Looking at log pasted above: ``` Jul 28 20:31:53 slc04 ERIGON[383187]: [INFO] [07-28|20:31:53.931] [Snapshots] Stat blocks=15143999 segments=15143999 indices=15143999 alloc=2.6GB sys=3.4GB ``` It seems highly suspicious that block 15143999 coincides exactly...

grepping a bit: ``` Jul 28 01:55:21 slc04 ERIGON[606667]: [INFO] [07-28|01:55:21.402] [snapshots] Merge done from=15130000 Jul 28 05:25:56 slc04 ERIGON[606667]: [INFO] [07-28|05:25:56.642] [snapshots] Retire Blocks range=15140k-15141k Jul 28 05:26:45 slc04...

Running with --snap.keepblocks did not prevent the issue from occuring

--snap.stop seems to be a successful workaround