Testing plan for the community

Hello Storj community

I have seen a few times now that whenever there’s traffic fluctuation there are old discussions popping back up discussing what had happened and generally just guessing what is going on. I wanted to ask if it is planned to have some kind of testing plan so the community know what kind of tests run during a specific period or when to expect test data and when not. I know we are not supposed to expect traffic but as an SNO it still is quite useful to know more insights and be prepared to some extent. Could also be useful to plan for maintenance and downtime if needed. It could also be used to verify with the community if expected usage is within what storj expects so bugs like the one last week could be found more easily

Tell me what you think about this idea

look into this thread :wink:

I would like it as well, but it’s not practical and could even be harmful.

Let me start with why it’s not practical. Storjlabs needs to be able to change testing patterns as needed to keep development going quickly. They may have a rough outline of what kind of test traffic or incentive traffic they would push on the network. But when they find something that needs attention right away, test traffic may have to be stopped immediately. Additionally, SNOs shouldn’t rely on this information being available in the first place as it definitely won’t be when traffic shifts more to pure customer data.

So why can it even be harmful?

That’s why. If SNOs wait until a specific time to start maintenance, it could mean many nodes are offline at the same time. This could lead in a drop in file availability, which would impact actual customers. You don’t want any coordinated behavior like that in a distributed platform. You should do maintenance when your node needs maintenance, not when the network is slow. Even if test traffic drops, customer traffic doesn’t follow that same pattern. If I were Storjlabs, this would be one of the main reasons to not communicate traffic drops in advance at all.

3 Likes

I monitor on my own, when traffic drops, that’s when i do monthly maintenance.

1 Like

its not unrealistic for us to get some warning as SNO’s when a new large customer may be coming. So if we are really testing the ability of the network to scale within a short period of time (let’s think months) for a large customer (let’s say someone using PB’s of storage).

We are already testing with this magnitude of data, but we aren’t yet really testing the SNO community’s response time. How big could StorJ network build in 3 months if we needed to? What is the scaleability? Are we relying on an influx of tons of new members sharing their extra space? Or are we expecting more of the increased capacity to come from old node operators adding additional storage to their long-term vetted (and thus prioritized) nodes.

What is the onboarding of new SNO learning curve delay?