How are future updates handled?

Recently a user had a request to rollback an update. He was told, he can’t. That made me think how future updates will be handled.

We all know from Microsoft that there are huge issues with Windows 10 updates from time to time. This poses 3 questions to me:

  1. How are updates deployed in the future to prevent SNOs to update to a faulty Tardigrade version at the same time causing their nodes to fail?
  2. What if a Tardigrade version turns out to be faulty and a SNO cannot rollback?

And 3. but this is more general: Windows 10 causes a lot of problems with updates. Would it be possible that a Windows update kills so many nodes that data is lost? Wouldn’t it be feasible to take into account for ingress data the OS of the node as well to distribute it fairly across differen versions?

1 Like

This is why phased rollout is used to begin with. To detect issues early.

In the past there have been some bugs that caused nodes to be disqualified and Storj fixed that for the effected nodes. I assume we would get similar treatment if it’s because of a software issue. I wouldn’t worry about that too much. Just make sure you patiently wait until it’s your nodes time to update. That lowers the chance of you getting a faulty update by a lot.


Think about that:

Every Windows Node gets updated slowly to the new Node-Version 6.5.4 (example).
Now every Node is updated to that version.
Maybe 1 week passes, everything runs fine and then comes Microsoft with Updates (which Storj can’t control how fast they get deployed) and in maybe 24 Hours the whole Windows Nodes shut down because of that Update.
You can’t predict that someone will find the cause in a short time.

How does the Business Continuity Plan looks for such scenario?

1 Like

Well that’s a different issue and it’s separate from how storj rolls out their updates.

Luckily windows feature updates use a much longer and slower phased rollout, so issues will be known long before it can damage the network. Monthly security patches are rolled out faster, but they are also much less likely to break something.

If something does break storj could release an intermediate update with a quick fix for that only. They could opt not to use the phased rollout for that fix.

What make things worse is that even if the os is a Linux, it could be a VM guest on a Windows 10 host. :stuck_out_tongue_closed_eyes:

It’s nothing new, you have always been forced to upgrade. Think of it like an online game. If I’m running a older version and you are running a newer version how can we game together when were have different code?