Still on v1.102.3 with 20TB node... wait?

Hey team,
I have two nodes, my smaller one with 6 TB is already on v1.104.5… and is much more responsive now, running on a small Synology.

The big 20TB is on a bigger DS1019+. Since these bloom filters were active you can see it takes longer and longer, this is 1 month view:


Now I thought just wait for it and let it run… should I continue to wait?
Feels like it is a lot of wear and tear like this.

What to do?

Updater spits this out all the time:

> 2024-05-18T14:05:19Z INFO Current binary version {"Process": "storagenode-**updater**", "Service": "storagenode-**updater**", "Version": "v1.102.3"}
> 2024-05-18T14:05:19Z INFO New version is being rolled out but hasn't made it to this node yet {"Process": "storagenode-**updater**", "Service": "storagenode-**updater**"}

The rollout of version is always staged, progressively updating nodes such that if there is an issue with an update, it will be caught before all nodes are updated.

To see the current progress of the rollout, you can checkout version.storj.io. Specifically, check the “cursor” section. That cursor is what decides what nodes should update (How to calculate update progress?).
In essence, the cursor will start at “0000…”, meaning no nodes should update, and end at “FFFF…”, when all nodes should update.
Any values in between will update a proportional amount of nodes. Currently the rollout is at “7851…”, which is close to 50% of the nodes.
The cursor slowly increases over time until its done.

If you cant wait, you could update it manually (How to force update to 1.104.1? (docker)), but I would not recommend it, especially given the migrations in the latest update. My best guess is that by some time next week the update should be rolled out to all nodes.

3 Likes

Patience grasshopper. This time next week you’ll be on the new stuff.

3 Likes

Did the rollout halt or slow down? My big one is still on v1.105.4 for a long long time and it really is wasting IO / usage on the disks.
The other one is on v1.108.3 already for quite a while and IO / etc. is much much better and stable / responsive.

The rollout is still progressing: it looks like 20-25% of nodes are on 108. And I know your node hasn’t been on 1.105 for more than a few weeks: so I don’t know what you mean by “a long long time”. :wink:

Haha yeah long long time is a question of perspective of course :wink:
What I mean is that the wear and tear is massive, also there was an update in between. So I had hoped for a bigger push (and I am a fan of not rolling out too fast) see the windows issue that occurred and was fixed quickly.
As said my main issue is on the huge IO it produces for “nothing” so to speak.

The 107 rollout was cancelled not long after it started: and they started upgrading people to 108 instead. So I’m sure that added a few days delay… but we’ll get there.

1 Like