Title says it all. The past two days, I’ve had about 180,000 to 190,000 “race lost” messages in my logs. This is on a Synology NAS DS1821+, 8 16TB drives in ext4 w/SHR. Been up 4 months, fully vetted.
I’ve started up a new node on an rpi 4 with a USB 8TB drive in ext4. It’s been up 19 days with 0 audits so far. I’m thinking about copying my NAS to an external drive, and running it on the RPI.
Even fast nodes lose some upload/download races: it’s more useful to compare how many requests you get vs. how many you win. There are scripts that will scrape your logs and show you the percentages, like this.
No, there are many reasons for a node to lose races, e.g. being far away from Storj clients, or your ISP not having good peering with them. But it is still a good idea to investigate this metric to know why you are losing races.
This seems to be normal these days. It was already discussed here.
In short: Audit should start after 1 month. Just wait.
Any customers behavior is normal. Your node cannot win all races anyway because it cannot be close to everyone customer in the world.
This is most unhelpful.
I don’t believe I ever said that I expect to win all races.
Ok. In short - doesn’t matter. If your node doesn’t have FATAL errors or I/o or database-related errors in your logs and your reputation (audit score, suspension score and online score are 100%) - everything works as expected.
The success rate is a relative thing - it checks only the latest logs, nothing else.
If you do not have failures for GET_AUDIT or GET_REPAIR - your node is working properly.
As a conclusion - the success rate usually do not show anything useful, unless the percentage is lower than 80%.
Thank you. This is helpful.
