Changelog v0.33.4

For Storagenodes

Update Process
Over the next week we will roll out the new storage node version. We will start with windows nodes because the windows updater gives us better control about how fast storage nodes are updating. We will start with a small percentage of the network. We will push the docker image later this time. Don’t panic if your storage node is not updating as fast as in the previous releases.

Used Space Graph on the Dashboard
We had problems with the used space calculation and the repair checker on the satellite side. Both services are running side by side with garbage collection. Garbage collection was running out of memory and the process restarted. The used space calculation and the repair checker didn’t finish. Short term we increased the memory. Mid term (this release) garbage collection will not run on the first iteration. It will start on the second iteration after the used space calculation and the repair checker finished once. Long term we will split garbage collection into multiple processes.
As a side effect, you should see a used space spike on the storage node dashboard because the used space calculation is compensating the gap.

100MB Free Space for Node Selection
Once per hour a storage node will communicate with all satellites and transfer some information like the available free space. As soon as the storage node has space for a single piece the satellite will start selecting it for uploads for a full hour. The storage node will accept a few pieces and for the rest of the hour reject all the other uploads with an “out of space” error. We don’t want that. Instead, the plan is that the satellite doesn’t select storage nodes with less than 100 MB free space (this release) and as soon as the storage node is falling below that magic line it should check in immediately and don’t wait one hour (planned for next release).

New Tardigrade Satellite

For Customers

10% Discount on STORJ Payments
If you pay $50 in STORJ tokens we will give you 10% on top. You will receive $55 on your account.

Known Issue with the 1 TB Coupon
In order to get the 1 TB coupon, you need to add a credit card or pay $50 in STORJ token (excluding discount). To trigger the coupon you also need to create a project.
The coupon is not getting added if you pay $50 STORJ token in 2 transactions. At the moment it has to be a single transaction with $50 or more. Please contact support if you are affected by this issue.

6 Likes

Don’t panic if your node is not updating.
Panic if it get updated too fast

2 Likes

usualy when changelog acure we can update, when update itself will come out?

You’ll see when it’ll hit

2 Likes

Monday according to the post

Now I’ll need to somehow remember to update my node on Monday.

If you’re on Windows, the update should happen by itself, and for docker CLI we don’t expect the update till later anyway, and watchtower will do it for you.

1 Like

Watchtower once failed to start the node after update and the update happened while I was sleeping. The result was 3 hours of completely preventable downtime.
After that I only update the node when I have at least an hour of free time immediately after to fix any problems that may be caused by the update.

3 Likes

I wonder if that is a common problem, my watchtower never has caused me any problems (so far).

Only problem ive seen was in linux watch tower wasnt updating the node probably a problem on my part though, But windows GUI I saw the issue of the update part of the GUI wasnt running so you had to start it manually.

It had something to do with configuration - IIRC my docker start command was incompatible with watchtower. It probably is fixed now, but lesson learned - automatic updates may happen when I am not available to fix whatever problem that can come up and with the strict uptime requirement (I know the checking is disabled now, but I might as well do everything as if it was enabled) I would rather not have downtime that can be easily prevented.

2 Likes

I’m hoping this update also contains the fix for the polluted logs caused by “aggressively close the connection after upload without waiting for the last acknowledgment from the storage node” talked about in: Context cancelled on all uploads/downloads

1 Like

The update mentioned in that topic is only a terminology change. It still won’t differentiate between cancelled transfers due to “losing the race” or “aggressively closing the connection”.

will update be announced :smiley: :smiley: :smiley:

Suposed to be pushed out on monday which today is monday. That was the announcement maybe they actually meant it was tuesday since today is a holiday. They also said windows will get the update first.

in america it start to be monday soon.

In which Timezone is the Storj team placed ?

that would be eastern US

So what is the time now

it is utc -5:00 hours