i have two nodes that only ipv6 the are working perfect on stefan benten sattelite but get no audits from the new sattelite wil the new sattelite also support ipv6
do you have data from new tattelite? no data nothing to audit.
no data no autid i can not find if the new sattlite is support ipv6
When is this planned for?
We don’t have a date for the shutdown.
Is it massive delete on stefen benten prepear for shutdown?
That question makes no sense. Lets say we are going to shutdown the satellite in 1h from now or 1 year from now. Was the action yesterday a preparation for that? Well you can answer with yes or no in both cases. As soon as the satellite is offline you can always argue that everything we did before was to prepare that final outcome. You can also argue that we could have taken the satellite offline without that step. Sorry I don’t know what I should answer to your question other then “unrelated question”
1 Like
I think you are missing the point of the question. Right now, if you were to shutdown the satellite, would all the data belonging to that satellite Be deleted automatically on SNOs. I think storagenode does not have that capability, so SNOs would be wasting there space. So the question is very valid and makes sense. While all technical capabilities are not implemented on storagenode side to remove a trusted satellite, there is a capability to delete all the data belonging to the satellite and free up SNOs drives
1 Like
That is a different questions. Yes we are going to clean up the space on all storage nodes. That way around it is valid. The other way around it is not. It doesn’t mean that all delete operations are a preparation for the shutdown!
Sure, but it was a question not statement, so you cannot say it did not make sense. You could answer: none of it was related to satellite removal, some of it was related, or all of it was related.
Oh now I get it. Ok so I have to select between “yes” “no” or “maybe”. I take “maybe” then. My plan was to avoid communication problems by explaining why the question makes no sense to me. That would give him the oportunity to rephrase it. It wouldn’t be the first time that I just miss understand the intention behind the question. But sure we can keep it short and answer the question with “maybe” instead.
1 Like
Let me rephrase it for you. As you may be already know all SNO like childesns. We need and like stabile envirament and know what will be next steps. If something not like in plann there will be panic and madness. Today we see that lot of people make panic, that no ingress, low egress and lot of files are deleted. And may be it is not directly your work to inform people about stuff going here. But it will be very nice if for example @jocelyn will inform people about some test plans for next week, once as week. It will hold people in peace and will make less panic. As someone told before that you do not have exact date of shutdown of stephenbente. May be resonable to make some around date like next month ot next 3 months. It is very needed information, as there is thousends of $ in held ammount, and ALL SNOs would like to know when we can get them back as sattelite will be shuted down. Is this explanation make more sence?
1 Like
That is the usual case when answering a question, Unless you are dealing with a robot.
If your worried about the held amount you could just run a graceful exit.
dont think this is good idea, as it can be shuted down in midle of it. And it deleteng the data now, lot of data.
Wouldn’t be the first time an engineer is accused of behaving like one. 
Keep in mind that there are sometimes language barriers in play that make things sound more snippy than they were meant to be. The way I read the response was that deleting data beforehand isn’t necessary and thus anything happening before the shut down isn’t really related to it.
This makes the question a bit like: “what is the color of happiness?”
Since emotions don’t have color and shutdowns of satellites don’t require deletions beforehand. Both questions don’t really make sense.
Additionally, while I would also appreciate any transparency about testing plans, it’s important to realize that they may also want to test things they haven’t planned. Run into issues that require new sorts of tests and plans may change at any time. Whenever you give more transparency and then things don’t go as communicated earlier, people still complain. So it’s a bit of a double edged sword. Especially since in the future, this information is mostly dependent on customer traffic patterns and it wouldn’t be possible to communicate this at that point anymore anyway.
4 Likes
All we know that if something is gonewrong is normal part of testing, but at least major plans cold be more transparent. Today we play here yet in sand box. There is very litle real world data here yet in compare with test data.
The difference between now and the future is that the aggregate patterns of many customers can be predicted, even though the actions of a single customer cannot.
Let’s take IPTV (unicast) for example. Suppose I have a server and a few thousand subscribers. The actions of a single subscriber cannot really be predicted - maybe today he won;t watch TV at all, maybe he will watch TV all day or maybe somewhere in between. Makes sense, right?
However, the aggregated actions of many subscribers form a very clear patter, similar to a sine wave with a “shelf” - very few people watch TV in the middle of the night, some people watch TV during the day, a lot of people watch TV in the evening. More people watch TV in the Sunday evening than any other time. If there is a basketball game or some other really popular broadcast, like the Eurovision, the traffic will spike to above the normal maximum as a lot of people will be watching TV then.
So, looking at the graph I can immediately notice some kinds of problems. If the traffic drops off sharply it means a problem - either a problem with the service or a bunch of people have no power. And so on.
So, I expect the same to be with Storj once it gets enough customers. It is extremely unlikely for 10000 different customers to agree to stop using the service at the same time etc.
Also, it can be possible to predict how much space my node will need, because, again, thousands of customers will not just agree to stop using the service or, conversely, to start uploading petabytes of data.
However, because currently the main traffic is tests, it is unpredictable. Storj can just stop whatever test is running right now and start a new test with vastly different traffic profile. Then they can just delete all files etc.
Still, if I see the traffic drop off sharply, I have to investigate, because the traffic drop may be because of some problem with my node - my IP changed and the DNS update script failed or something else.
For example - before the deletes started, the used space on my node would increase by about 10GB per hour. I had calculated that I had enough space for a few more weeks and should start looking for new hard drives. If all of that data belonged to actual customers it would be easy to decide to expand my node, after all, the customers do not care or even know how much data exists on the network and are not going to stop uploading data just because the network is getting full. On the other hand, Storj does know how much data is on the network and can stop at any time. And yes, the deletes started before I bought new drives, postponing that.
4 Likes
Yeah, I don’t disagree with any of what you said. But I think there is a chasm between customer traffic becoming the larger part of traffic and eventually having thousands of very active customers so that it evens out. Customer behavior will probably take over whenever there are a few large customers with completely unpredictable behavior. So we’ll likely have to deal with unpredictability for quite a while before we get to that evening out across thousands different customers.
1 Like
I guess it depends on what those customers do since that may be predictable (for example, using the network as a CDN - then the load will depend on how many users access it and there would be many users etc). And even if there were only a few large customers with completely unpredictable traffic, the result would still be a bit more even since those customers most likely won’t be coordinating between themselves.