Problem: Low TPS Current Workaround: Periodic Server Restarts My sugestion: A mix of the current workaround and this: If TPS is <= x for more than t time do an auto server restart [x=minimum bareable tps value] [t=time (example 5 to 10 min)] that way, if the server starts having very low tps and the periodic server restart is yet to come, this would fix it (i think?) I hope this might help
I see how you think this may work, but let me show you a situation where it will not. A few days ago, I had to use WorldEdit to remove tanks on infinity node 1, which single-handedly dropped tps down to ~0.5-2. The tanks by themselves weren't so hard on the server, it was just the amount that was placed in the same area. A shocking ~1500 tanks were in the same area, and with each one being minimally bad for the tps, the accumulated effect made everyone time out. Now, if your solution was put in place, then the server would have just been put on a restart as soon as the owner of this setup logged in. It would have been rather bothersome. We co-admins are here to fix tps issues. If you ever feel that a certain server is having issues, just please holler to nearest staff-member, and if they are not co-admin, then they will contact us in staff-chat.
I see your point and it seems the nature of the tps drops is very unpredictable. Anyways its still a good thing we have you guys to fix the servers up if they need manual work keep it up
I see where you come from and have thought similar. The main issue I see is that a system like this probably lengthen the time until it is being looked into. All 1.7 servers have a bug between cauldron and forge where on rare occasions an atomic list is going crazy. This happens probably 2-3 times a day at different times. The auto restart would do a lot good in this particular case, but what if it is something else? What about lag spikes instead of constant lag.. of course the system can be extended and teached to detect those but in the end we might end up putting all the time in work around's instead of the initial issue. As much good as technical automation can do, once the farmer doesn't know the names of his cows anymore he doesn't know when something is wrong with them. If we ever implement something like this I would want to leave the final decision to human thinking, so if no staff is online, a possible restart vote is possible with tps below 10 for example. In the past I built a site keeping track of all the servers and even speaking up when tps are low or the server offline for long time. This tool is open in background all the time and alerts once something is wrong and leaving the work to us. This allows us to look into the issue and see if it is a known thing or something new and maybe there is even a simple solution which gets rid of the particular issue on the long run. I'm around +12 hours a day, SirWill is around +12 hours day and the other Admins with direct server access are around for at least 4+ hours a day. While we don't have shifts most of the time always someone with full power is online.
