This is probably months late, but I logged in to see all ten of my servers at a perfect 20. I’m glad to see that the issue was resolved. Great work Ask!
It’s still happening for some operators, but the changes definitely did smooth it out.
I have to give a big thanks to @stevesommars who’s done (and continues to do) a huge amount of diagnostics and trawling through tcpdump and traceroute data collected on the monitors and from collaborating server operators.
His work paved a path for making a simple change that helped without (or with minimal) negative side effects.
Hey-ho. Good to hear @ask that it has improved for most people/servers, and, at the end of the day, it is a pool and therefore generally good news. For me in the UK, on FTTP 1Gbps enterprise symmetric static feed, not so good since Wednesday.
Having checked and rebuilt everything when this happened before, I have learnt that the thing to do is - ‘sit-on-my-hands’. I’ll leave it for a coupe of weeks, hopefully it will improve, or may as well pull it from the pool.
All the best,
I added my 4 servers again.
Let’s see if it did improve.
Steve does monitor my servers for almost 6 months now and continues to do so.
Let’s hope it worked out…fingers crossed as I really want to help to keep the pool in shape.
Don’t know what you did, but the fix of the monitor works really well.
I see in the logs that it still misses from time to time but mostly connects just fine.
It’s not a perfect 20, that doesn’t matter.
I’m very happy it doesn’t drop a server instantly and starts mailing.
Back in the pool for a week now, not 1 single mail or drop below 0.
Thank you very much for fixing and/or changing the algorithm!
Just logging into the server manager now, each of my 11 entries are a 20 score. I recently added my montreal server back and it was only doing about 250qps at the 100Mbit setting in the zones it’s in. I’m going to work at getting that one higly productive again. I may bump up tick.nc.us.logiplex.net because it has been a perfect 20 for numerous months at the 250Mbit setting, but I will compare it’s current qps with its bandwidth allotment and let you guys know here. The other 7 will continue to do exactly what they are right now. The primary logiplex.net server out of the six is at 132 days uptime right now which it shows on the website.
Again, excellent work with the monitor Ask.
I was so happy and then it happens again.
Newark is having timeouts and the almost good score is gone again.
So far I got no email, that is a good thing. Fingers crossed.