Cardano-db-sync: Offline pool metadata fetch: fetch errors

Hi all, I recently got db-sync running and am attempting to populate the database using:

PGPASSFILE=config/pgpass-mainnet db-sync-node/bin/cardano-db-sync \
    --config config/mainnet-config.yaml \
    --socket-path $CNODE_HOME/sockets/node0.socket \
    --state-dir ledger-state/mainnet \
    --schema-dir schema/

I keep getting “Offline pool metadata fetch: 43 results, 20 fetch errors”

[db-sync-node:Info:79] [2022-03-14 19:44:20.57 UTC] Starting epoch 276
[db-sync-node:Info:79] [2022-03-14 19:44:22.19 UTC] Handling 667460 stakes for epoch 276 slot 33868855, hash 637539836f498afdd2c69fbf9bf47207d7388fbe6eb23e56030e544d109524ff
[db-sync-node:Info:79] [2022-03-14 19:44:33.37 UTC] validateEpochRewards: total rewards that become spendable in epoch 276 is 15878394.99255 ADA
[db-sync-node:Info:79] [2022-03-14 19:48:23.04 UTC] insertShelleyBlock(Mary): epoch 276, slot 33868855, block 5937586, hash 637539836f498afdd2c69fbf9bf47207d7388fbe6eb23e56030e544d109524ff
[db-sync-node:Info:79] [2022-03-14 19:48:25.23 UTC] epochPluginInsertBlockDetails: epoch 275
[db-sync-node:Info:79] [2022-03-14 19:49:32.96 UTC] Offline pool metadata fetch: 43 results, 20 fetch errors
[db-sync-node:Info:79] [2022-03-14 19:50:13.15 UTC] insertEpochInterleaved: Epoch 276, 667460 stake addresses
[db-sync-node:Info:79] [2022-03-14 19:50:33.09 UTC] Took a ledger snapshot at ledger-state/mainnet/33886647-9d7316d45a.lstate
[db-sync-node:Info:79] [2022-03-14 19:50:33.09 UTC] Removing valid files ["ledger-state/mainnet/33856796-59cf134b9d.lstate"]

Searching a bit I found people with similar issues (but not exactly the same) back in 2020-2021… anyone know what might be causing this?

System specs:
8 cores, 32GB ram, 256GB SSD

Also FWIW, the first ~260 epochs didn’t have this issue.

Nothing out of ordinary, you’d expect these as it has to grab all pool’s metadata information - which if off-chain, and some are not reachable/inaccurate. You can ignore those wrt wait times - note that by default dbsync isnt very chatty in logging, so you’re not expected to see a lot of movement in log every few minutes.

System specs:
8 cores, 32GB ram, 256GB SSD

[ Unrelated to your query] This would be minimum resources for dbsync instance by itself, but you’ll likely run out of resources with Postgres DB and node depending on your server tuning , I’d highly advise you to bump up as it’s not gonna be sustainable regardless even if you use workarounds (swapping, throttling, etc)

ah that makes sense. thanks for that.

what would you recommend (ideally) in terms of cores, ram, and storage?