Changelog v1.35.3

For Storage Nodes

Multinode Dashboard Binaries
In the last release, we added some multinode dashboard test binaries. With your help, we have been able to prove that these binaries are working. Thank you very much. From now on the binaries will be included in every release.

Audit Score Safety Check
A few weeks ago we noticed that the audit score on the us2 satellite was decreasing. The root cause was a bunch of files with an expiration date. All storage nodes have deleted these pieces as intended. The satellite was still sending an audit. We managed to fix that but we also decided to implement a safety check at the end of the audit. If not enough (29) storage nodes have passed the audit we don’t decrease the audit score assuming that the failure is on our side and not on the storage node side.


That is very clever, this will prevent basically all possible faulty audits from impacting the score. Though am I right to think the node will still log this error on its end?


Yes. It if you see such an failed audit please still escalate it so that we can find the root cause and fix it. This safety check is only there to isolate the consequences of that bug but we would still like to fix it.


That’s what I thought, that’s perfect actually as it means issues won’t go unnoticed, but effects are mitigated. I don’t think you have to tell people to keep reporting faulty audit failures, that’ll definitely happen either way. :wink:


hope for success, but always try plan for failure… :+1:


Wow, that must have been a scary sight on your end! In normal circumstances this would mean the network lost a piece.

I wonder, what was the closest the network was to actually losing a piece?

That’s actually shown in the stats made available.

1 Like

This is current piece data, not historical.


No staged rollout this time? Docker image is already available. 2 of my nodes already updated.

There was a staged rollout, I noticed it yesterday. It has since finished. I’m guessing there were only minor changes to storagenode code though.

The update was pushed on github 5 days ago :slight_smile:


Would I be correct in assuming that there will be an indication in the changelogs if the MND binary has changed and needs to be updated?

I would recommend to update it every time for simplicity.

Are there any plans to make MND available as a docker image so we can update it with watchtower? (And simplify installation on more restrictive systems like NAS systems?)


7 posts were split to a new topic: Online score just drop down…how to get 100% again?