The old reputation system tracked audits and uptime checks for the entire lifetime of a storage node. This didn’t allow us to disqualify nodes quickly enough, nor did it allow nodes to recover from a bad period in their history. The new reputation system gives recent events like audits and uptime checks a higher value in order to represent a more accurate reputation for the node. https://github.com/storj/storj/blob/master/docs/design/node-selection.md
The node disqualification system has been updated to use the new reputation model.
This means a disqualified storage node can’t get qualified again, even if its reputation goes above our threshold. Disqualified storage nodes won’t get paid and won’t get selected for uploads or downloads. https://github.com/storj/storj/blob/master/docs/design/disqualification.md
We added signed vouchers from satellites to storage nodes, this will be used in the future so that storage nodes trust each other. The vouchers have an expiration date, and if a node gets disqualified the satellite will no longer renew that voucher. ttps://github.com/storj/storj/blob/master/docs/design/kademlia_audit_gating.md
We fixed an edge case with our audit service. There was a small chance the audit service would select a segment while the customer was deleting the file. In this case, the audit would fail. The auditor will now check to see if the segment was deleted before counting it as an audit failure.
We fixed a bug that was causing orders to not be accepted properly because of a data wipe. We restored the database from a backup on US-Central-1 and Asia-East-1 Satellites. The fix involved the Satellite rejecting “invalid” orders so the valid ones can be processed.
We deleted all buckets and files from Tardigrade Satellites(US-Central-1, Europe-West-1, Asia-East-1). When the Storage Node is updated it will delete the data from those three Satellites. The data on Europe-Central-1 wasn’t touched.
Satellite
We added IP filtering in our node selection process so operators who are running multiple nodes aren’t incentivized based on that criteria.
We renamed the Tardigrade Satellites from Mars, Jupiter, and Saturn to US-Central-1, Europe-West-1, Asia-East-1. This new naming convention will make it easier for clients to choose which satellite best suits them based on location.
Uplink
All uplinks are REQUIRED to update their binaries to the latest version to avoid compatibility issues.
We wiped all of the client’s data on the Tardigrade Satellites.
We completed the C bindings for the libuplink library.
Thanks! Turns out my Watchtower docker image had not started in some time and was out of date. I’ve pulled the latest and implemented this for updates moving forward. I’ve also documented this all so that I can verify it next time. Appreciate your time on this.
Thanks, I had it installed but somewhere along the line it stopped and was an older version. I updated it and then reimplemented it to update storj so hopefully next time this will all work seamlessly.
I using it on Debian/Ubuntu Server LTS, but it doesn’t matter, watchtower is just docker container and will work on any docker host (synology too).
Also my watchtower setup is unusual, I use email alerting (when update is coming I got email).
I’ve done the manual update (stop storagenode, rm and then pull /storjlabs/storagenode:alpha) as well as install the watchtower container using the script in this thread, however the node still shows as running 0.14.11. Any ideas?