@kevink thanks will update the dashboard, can you point me to where onlineScore change was discussed please? Has anything else changed about it? Totally missed it. I added a metric for onlineScore api value but I was sure the old one still works and both are the same thing.
This is what I found:
There is definitely more but not about the API itself: Blueprint: tracking downtime with audits
1 Like
That helps thanks. What do you think would be reasonable thresholds for this and maybe audit as well? Currently I got 99.8 - orange | 99.5 - red text colour thresholds set for these cells.
That’s merged to master here for both summary and sat tables https://github.com/anclrii/Storj-Exporter-dashboard/blob/53dee2e8ac653d898f840b2745e8ad26ab86e3f7/Storj-Exporter-Boom-Table.json
I’m not quite sure either. I used red below 97%.
Currently you can have a downtime of up to 288h per months (which will change at some point) so that would be 60%. But making everything below 60% red seems really low.
But 99.5% is definitely too high. That was a very old requirement that won’t be enforced again so I wouldn’t go that high for the red threshold.
Hi,
What is the Metric code thingy for Grafana to get the storj uptime percentage? Mine still shows as 0%
It should work with latest version of exporter and dashboard. Have you updated both and still having issues?
How do I check my current version of the exporter?
I have seemed to fix the issue, for some reason my storj-exporter image wasn’t updating (I have to build it manually), but a docker prune fixed it.
Just curious if there’s any reason to run my own grafana instead of using the free cloud based one?
I don’t need more than one user, and I am not aware of more than 5 dashboards for storj?
Thanks and hope you are well.
edit - Also why dont we add the shared dashboards on there?
I have both but keep using the cloud one since I tried it. So far so good, as long as limitations work for you I don’t see any problems with it.
Not sure what you mean, add where?
My question would rather be: Why would you use a cloud grafana if you are deploying your prometheus locally anyway?
In the cloud grafana there’s a ‘shared dashboards’ where storj could have a dashboard that we all use so that we dont have to individually update all the time.’
This isn’t relevant to me as I have machines on various networks/locations so no matter where I put the grafana, it will never be local to all nodes.
Ok I see what you mean, will look into it, good idea.
2 Likes
I wish we could add this into the same docker i’m running for each storage node. Having tons of extra exporter dockers is annoying. I guess modifying storj’s stuf though would mean the updater wouldn’t work right.
Getting this setup today though and testing out that free grafana to view it all at.
1 Like
you could build your own image from both sources… probs not rcommended though 
Ideally it would be great if storj binary had the functionality to expose prometheus metrics. This is quite a common practice when building modern apps and when I started with storj I was expecting it to be there. Asking around I was pointed to a message that mentions prometheus metrics are exposed in storj debug mode but I couldn’t find it there either… so I made my own and sharing it in case it’s useful for others. I considered contributing to storj codebase directly but I figured it would take a while to negotiate and I have no experience with golang, my programming skill is very basic and python was a quick solution for me.
Since 1.0.1 release it seems to be fairly stable and fail-safe and maintenance for any upcoming changes in api should be easy enough to support it further. I was going to start looking at contributing to storj at this point but with current traffic rates I’m quickly loosing interest. I’d probably be out by now if I didn’t have the exporter to support (my earnings don’t cover my running costs currently 
). Though I’m still happy to support the existing exporter implementation for as long as people use it.
I’m not sure if there’s a better alternative to running it in a second container, I’d stay away from messing with the official storagenode images. There’s an option to run exporter as a systemd service if that’s more convenient, but docker works great for me so far. I’m not a fan of aggregating multiple exporters into one exporter container or tracking multiple storagenodes with one exporter. Happy to discuss other alternatives on how to improve this side of things.
1 Like
Thanks a lot man you rock! If we ever cross paths, the beer is on me.
How can I find Docker container hostname or FQDN to fill - STORJ_HOST_ADDRESS ? Thank you.