Server health service

[Nolokor]

Please consider implementing service which monitors server status and restarts server automatically.
If you need debuginfo I guess there's some solutions for that too in Java to create a crashdump of your crashed process.
It improves service quality quite a lot.

[boshaw]

Please consider implementing service which monitors server status and restarts server automatically.
If you need debuginfo I guess there's some solutions for that too in Java to create a crashdump of your crashed process.
It improves service quality quite a lot.

The server isn't in java.

[Nolokor]

The server isn't in java.

If it is a native executable, then it's even easier as in linux there's tools to save crashdumps automatically. So you don't need to run it under gdb or smth and you can just add notifications (email alerts?) and debug those dump files post-factum, while server will experience only brief downtime.

[Ysh]

The server isn't in java.

If it is a native executable, then it's even easier as in linux there's tools to save crashdumps automatically. So you don't need to run it under gdb or smth and you can just add notifications (email alerts?) and debug those dump files post-factum, while server will experience only brief downtime.

Maybe there is already system like this in place, but it has failed somehow!

[MagicManICT]

This absolutely cannot be done for multiple reasons. Please research out your thoughts before posting in C&I. This has been hashed out multiple times over the years. Thanks. (And folks with more experience on the matter has given better technical explanations.)

1. Server code is all C
2. It's a development system, not a live production system
3. It's an unstable system, clearly labeled as "alpha grade" software.

Crashes can leave a lot of dangling data out there that can become lost or even corrupt the entire system if automatically restarted. A LOT, and I cannot stress "LOT" enough, goes into creating such systems for production servers along with tons of redundancy and back-ups that allow for things such as this. Maybe if Haven ever becomes 'that' stable, then such a system can be developed, but that's likely going to mean it has grown significantly beyond the means of two people to design and develop it, too.

What if the data and process that caused the crash is in a "delayed reaction" meaning that it got saved and is present when the server restarts? Now your server is going down every few minutes and restarting and it has not a clue that the problem state is still active in the system. Now, after 6-8 hours of this (a normal person's sleep time), the world is totally corrupt. Now you've just had 6-8 hours of downtime, and possibly down all day to restore from backups that might be days or weeks old (depends on how data backups and database transactions are recorded).

Account time suspends during downtime. characters don't progress, account subscription time, etc. isn't used up.

[Ejnekor]

Account time suspends during downtime.

While this is true, I beg to differ. What if someone went into hunting trip on a wagon? Well, even a normal boat can be precious at times. What if just before the crash person opened gates to exit/enter and cannot be online for the whole duration of downtime like this? His whole property now in danger because he won't be able to return and close the gates now. I am sleepy as hell and waiting for the server to come back on because I am in somewhat similar situation to this and wouldn't want to have anything bad to happen while I will be away once they come back on.

[MagicManICT]

What if just before the crash person opened gates to exit/enter and cannot be online for the whole duration of downtime like this? His whole property now in danger because he won't be able to return and close the gates now.

this is why you always have a visitor gate to exit through. If it gets stuck open... eh, no big deal (short of some bug someone might try to exploit).

Again, it's an "alpha grade" piece of software. Bugs and downtime come with the territory. While there's not express consent forms you digitally sign when you make an account, this is clearly presented.

[bmjclark]

Account time suspends during downtime. characters don't progress, account subscription time, etc. isn't used up.

Characters with outlaw stay logged in though. Meaning if the server goes down for 8 hours and my char is outside, and i cant be on when loftar gets out of bed or home from wherever to do it, i could very well be dead. Im not advocating for the OP's suggestion, but there are certainly shitty things that can happen to you because of extended downtimes.

[walter]

Loftar prefers to sysadmin as if it were the 90s. It's his hobby... if thats the way he wants it, what can we do about it.

[shubla]

Loftar wants to fix bugs first before restarting the server.
No point in automatic restart if it will just crash again.