DayZ Restart Schedule

DayZ dedicated servers get worse the longer the process runs. Entities pile up, a script leaks, a mod holds a handle, and the simulation that felt fine at noon feels late by midnight. A restart is the unglamorous fix: stop the process, start it again, keep the hive and the persistence. It is not a wipe. If you call it a wipe, players will log in expecting empty tents and will find their base, then decide you lie about other things too.

Pick a clock and keep it. Every six hours, or every eight, or once a night if your mod set is light and your players are in one timezone. The right interval is the one you can announce and the one the server actually needs. Watch the player chart on your DayZ Manager page. A cliff that is not on the schedule is a crash. A cliff that is on the schedule is a restart people can plan around. Put the times in the hostname only if you will update the hostname when the times change. Stale "restarts 00/06/12/18" text is a broken clock.

Announce with say before you do it. Ten minutes, five, one. Then restart. CFTools Cloud can own that sequence so it does not depend on a moderator who went to make coffee. A scheduled restart that sometimes happens is worse than a manual one, because players stop believing the message. If the scheduler is on, leave it on. If you need to skip one for an event, say so in the same channel you usually say nothing, because the schedule was the channel.

Restart with the process supervisor you trust. If CFTools Architect is the host, the process comes back because that is the host's job. If you are on a VPS, a scheduled task that starts the binary is the difference between a restart and an outage. Test it once in the afternoon. A task you wrote at 3 a.m. and never saw succeed will fail on Saturday in the way that empties the server.

Do not restart to apply a loot change without reading the loot change. A bad types.xml will crash-loop and your clean schedule becomes an outage. Edit, restart once while you are watching, confirm you can join, then let the clock take over. The loot guide is the edit. The startup line is what actually launches. Cloud sends the restart command. Something on the box has to be willing to boot.

Players will complain about restarts no matter what you pick. They will complain more about desync and about a server that has been up for four days and feels like it. Tell them the reason in one sentence and do not debate it in global chat every cycle. The schedule is the policy. Exceptions are rare and written down. If you are also planning a wipe, do not stack it on a surprise restart. Wipe at the announced time, on a stopped server, as its own act.

After the restart, glance at the public list. The server should come back online, on the same map, with the same mods. If the rank page shows it offline for a long time, the process did not come back. That is a host problem, not a messaging problem. Fix the boot before you add another announcement.

Watch one cycle before you trust the clock

Pick the minute from the player chart, not from a template. If your country is busy at 21:00, do not restart at 21:00 because a guide said "evening is fine." Restart in the trough, and say the timezone out loud. UTC in the hostname and local time in Discord, both correct, is the whole courtesy. A hostname that still says 00/06/12/18 after you moved the clock is a lie the server list will display without checking.

Run the cycle once while you are in the mission. You should see the say, the disconnect, the process return, and your character where persistence says it should be. If the character is fresh, you did not restart. You wiped, or you booted a different profile. Stop and read the paths before the scheduler repeats the mistake every six hours. CFTools Cloud can fire the command on time and still point at the wrong RCon port if you cloned the server and copied the job. One job per process.

Skip the restart only for a reason you announce. A silent skip teaches people the message is decorative. A double restart, because the host panel and Cloud both fired, teaches them the clock is hostile. Disable one. If CFTools Architect supervises the process, let Cloud own the announcement and the command, and let Architect own the boot. Two supervisors racing is an outage with a log that looks like success.

After a week, compare the chart to the policy. Cliffs that are not on the policy are crashes. A modded server that crashes ten minutes after every boot is a mod, not a schedule problem. Lengthening the interval will not heal it. Fix the boot, then put the interval back. Players will complain about any restart you choose. They complain less when the same minute happens every day and the base is still there. That is the entire design. A wipe is a different announcement, on a different day, with the process stopped for a file delete you meant.