SpeedBump Hit on a new account: requesting a higher node limit (org_5fVi0IXjUr16TWP6)

[virtusapps]

Hi team,

New account here, hitting SpeedBump Hit on every server pool and stuck at 6 nodes total. I found the explanation in #198 (@sirishkr: "We have speed restrictions in place that prevent new users from coming in and bidding for excessive amounts of capacity"), so I assume this is the new-account restriction rather than a market/capacity issue. I'd like to request a higher limit.

Org: org_5fVi0IXjUr16TWP6

What we see (us-east-iad-1, cloudspace channels-veo-render):

Server class	Bid	Market price	desired	wonCount	Status
ch.vs1.2xlarge-iad	$0.160	~$0.035	6	1	SpeedBump Hit
ch.vs1.xlarge-iad	$0.100	~$0.005	10	2	SpeedBump Hit
ch.vs1.large-iad	$0.080	~$0.001	20	3	SpeedBump Hit

Stuck at those wonCount values for over 2 hours. Bids are well above market, so this does not look like losing the auction.

We also tested other regions, in case it was per-cloudspace: fresh cloudspaces in us-central-dfw-1 and us-central-ord-1, each asking for only 6 nodes, both went straight to SpeedBump Hit with wonCount 0. So it appears to be account-wide. Those two have since been deleted.

Two things that would help us a lot:

1. What is the actual limit for a new account, and what makes it lift over time? Nothing about SpeedBump appears in the docs, so we've been guessing at the mechanics.
2. Can the limit be raised for our org? We're happy to prepay or provide whatever information you need.

Our use case: batch video rendering. Work arrives as a queue of independent chunks, so it parallelizes almost perfectly. We create a cloudspace, drain the queue, and destroy it, typically a few hours per day. We'd like to reach roughly 30 to 40 nodes of ch.vs1.large / ch.vs1.xlarge during those windows. Cost is not the constraint; the hourly bill is the same whether we use 6 nodes for 5 hours or 30 for 1 hour. What we need is the wall clock.

One more question on the mechanics: does the restriction trigger on the size of the jump in desired, or on the absolute number of nodes? We first ran with the native nodepool autoscaler and reached 16 nodes without any problem. We later switched to setting desired_server_count ourselves (we needed three server classes scaling in parallel, and only one pool per cloudspace may have autoscaling enabled), and raised desired from 6 to 31 in one step, which is when we first saw SpeedBump Hit. If the trigger is the jump rather than the total, we'll happily ramp gradually, but we'd rather follow a documented rule than guess.

Thanks!

[vjha-pf9]

@virtusapps I’ve removed the speedbump limit from your account. Please check.

[virtusapps]

Thanks

[sirishkr]

@virtusapps welcome to Spot! We are happy to have you!