Open andy108369 opened 2 years ago
Based on the above, it makes me think that too many existing bids (20): unknown provider
message is misleading. As there is no such issue when running the akash-provider against the archive RPC node.
I haven't seen this issue before, is this blocking anything in prod?
it is a network parameter that limits to 20 bids for one order https://github.com/ovrclk/akash/blob/7fba1c7b05a37d7d043411a423e383e6c861d10f/x/market/handler/server.go#L42
Thanks for the context. It sounds like the provider should avoid trying to submit a transaction for a new bid in that scenario.
I've tried limiting deployment to the specific provider using placement group attributes, as well as running the state-sync RPC node, got the bid without issues.
placement:
akash:
attributes:
host: akash
datacenter: equinix-metal-ams1
Have noticed lots of too many existing bids (20)
on the other providers today.
I think it's time that we lift the OrderMaxBids
from 20
to 40
or 100
.
$ akash query params subspace market OrderMaxBids -o json | jq
{
"subspace": "market",
"key": "OrderMaxBids",
"value": "20"
}
Related code paths:
Moved OrderMaxBids part to => https://github.com/ovrclk/akash/issues/1662
I'll rename this issue to "provider should avoid submitting a transaction when it hits OrderMaxBids".
Additionally, the unknown provider
message in the error too many existing bids (20): unknown provider
needs to be changed.
Perhaps ErrUnknownProvider indicates an invalid chain parameter
& ErrInvalidBid indicates an invalid chain parameter
comments as well.
As per @tidrolpolelsef, looks to be a copy-paste error: https://github.com/ovrclk/akash/blob/7fba1c7b05a37d7d043411a423e383e6c861d10f/x/market/types/v1beta2/errors.go#L106
akash v0.16.4.
Akash Provider hits
too many existing bids (20): unknown provider
errors when is using a state-sync RPC node, but is working fine when we've switched it back to the primary (archive RPC node):This transaction is
failed to execute message; message index: 0: too many existing bids (20): unknown provider
https://www.mintscan.io/akash/txs/CE5C3C45D85BC154918E3EBE8B6DFEF235B701485BFC2F3DCE6E4F04B0D35AB2Haven't tried the non state-sync & non archive RPC node (i.e. the snapshot RPC node).
Seeing
unknown lease for bid: invalid request
in the logs, I think it could be somehow related to that the AMS1 host is running older deployments which state-sync RPC node doesn't have indexed, leading to this sort of issue.