I guess there’s no mercy for all of us that were impacted by the bomb cyclone a week ago.
Power was back at my place after 6.5 days, but Internet is still out.
Given the climate change, the expected outcome is that some disruptions will take longer to repair for utilities.
My first node was only 1 month old, so there not so much to loose.
Has StorJ preparing the network for such massive disruptions in the future?
Is there a way to communicate with the network during an outage by using a high latency, low throughput network (such as a cellular hotspot) to let the network know that node hasn’t permanently vanished but is healthy and not yet ready to accept new writes yet?
Just bring your node online as soon as possible. It could be disqualified if it would be offline for more than 30 days. Before that the downside is that your node is losing data because of repairs.
You can communicate with the network though any internet connection, no matter the bandwith/speed. If the connection is slow, you loose races, aka upload/download less data. The satellites audit your node many times a day, to see if it is online and the pieces are healty, so you should keep it online as much as possible. After 4h offline, the audited/requested pieces are considered lost from your node, and are recreated on another node. When you come back online, your node will receive bloom filters that will move those pieces to trash. So after 4h offline, you start to loose data. As online score, only after 30days offline your node is DQed and you can delete it and start over with new identity and new node.
In your particular case, with cell connection, you can limit/block the ingress by tweeking the allocated space to a minimum, and reducing the minimum allowed, which is 500GB if I’m not mistaken. Maybe you can set both to 1GB, I don’t know what are the extreme limits.
You can’t limit the egress, because there are audits too, so if you block the egress, you are basicaly offline from Satellites point of view.
You could tweek the concurrent connections parameter, to limit it somehow. Or to use QoS bandwith limiter in your router’s config.
Maybe others have more ideeas.
Great suggestion! However, I wonder if every connection type will work. Cellular connections often use shared IPs (via CGNAT), which might make them unsuitable for connecting to the Storj network. Using a VPN with a dedicated IP might be a good way to overcome this limitation, though.
I didn’t say you can use any connection as a long term solution. If you want your node to perform well and make you some money, you should have a good connection, meaning unlimited data plan, small latency, good speeds.
But as a temporary solution, you just have to be online, no matter what.
Yes, you need a public IP or use a DDNS service to get a public address, that will be asociated by the service with your new IP, when it changes.
There are many discussions here about CGNAT cases also.
Xfinity finally restored the service - it was down for 1 week and 1 hour. The node was back online and I updated its DNS entry. After a restart, the status is green and I can see many successful GET and PUT requests per second. However, the status remains suspended.
Am I worried too soon? The node has been online for about an hour.
Nope. Your node need to have an online score greater than 60% to be unsuspended.
The online score should fully recover after 30 days online. Each downtime requires additional 30 days online to recover.
Thank you, Alexey! I appreciate the prompt response. The online score is greater than 70% for every satellite, but the status still displayed as suspended:
Your node shouldn’t be suspended with this online score. So, please, keep it online.
However, your node would lose some data while it was offline anyway, since it was longer than 4h, as explained earlier.
Well, my node was disqualified less than 24 hours after I brought it online and got “Back Online” emails from all four hosts. It’s confusing because it started accepting traffic and looked normal.
I appreciate if someone can investigate this issue:
The cyclone wasn’t a massive disruption for the Internet, so it wasn’t an issue: as there were always tens of thousands of nodes online. Even Russia could disappear and Storj would be fine. Customer files are safe!
The disqualification is happening when the audit score is below 96%.
So, either your node was offline more than 30 days, or there is a data corruption or lost.
See