Scheduled Server Reboots



  • I make everyone here do manual for all clients. We have seen too many hang ups etc.



  • I have been doing it with vSphere for a while now without issue.



  • @NetworkNerd said:

    For those that have a dedicated green zone for weekly reboots

    I definitely need to implement one of those.

    I'm thinking a batch job plus an e-mail when the server has rebooted. That way, if you don't get an e-mail, you know there is a problem (eg hanging) and can sort it out.

    Is there something about automation that makes hanging more likely or something? Or do you just mean it hangs anyway and it's better that you see it at the time rather than on Monday morning when everyone is yelling?



  • I'll agree with Minion Queen, I prefer to be available when my servers reboot. I've had more times that I care to remember when it won't reboot for whatever reason.

    Now that said... I've never (where is some wood to knock on) not had a VM guest boot correctly outside a bad patch.



  • How often do you do reboots, Dash? And can I call you Dash?



  • I use PDQ Deploy and schedule some servers. Others I do manually.



  • @Carnival-Boy said:

    How often do you do reboots, Dash? And can I call you Dash?

    he likes dashey....from what I hear.



  • Manually is great when you have a handful of servers but with anything more it's too much to do manually. Some network monitors with email alerts could help with whether the comeback up or not. I've personally never rebooted the VM Host itself while I'm not there though.

    Also windows server 2012/2012 R2 is a lot better with handling updates even if they fail. I have not had a single one of those not boot after an update sometimes they will delay and have to uninstall and take a little longer.

    I personally have also used an extra VOSE instance and use it for testing the server updates manually before pushing them out to make sure everything will work fine. I'm also usually a 30-45min early to work kinda of guy so I could catch something like that if it was down before anyone else got there.



  • When you inherit sketchy junk it takes a good long while to build trust. Unless I've built it or am intimately familiar with it's quibbles and quirks, no way am I going to trust a scheduled reboot.



  • @Carnival-Boy said:

    How often do you do reboots, Dash?

    Only if the server has a problem or if Windows updates requires it.

    Why should I do it more often than that? Linux boxes have been known to go years without rebooting.

    I have a server that I think would go years if not for Windows updates...

    And can I call you Dash?

    Yes Dash is fine. It's what I give restaurants, no chance of a name confusion with this one.



  • @Minion-Queen said:

    I make everyone here do manual for all clients. We have seen too many hang ups etc.

    On servers? That's part of the purpose of reboots is to catch systems that don't come back up on their own.



  • @Dashrender said:

    I'll agree with Minion Queen, I prefer to be available when my servers reboot. I've had more times that I care to remember when it won't reboot for whatever reason.

    I agree, but being available doesn't mean you do it manually. Scheduled reboots plus good monitoring and ta da.... less overhead, better coverage and you are still available.



  • @MattSpeller said:

    When you inherit sketchy junk it takes a good long while to build trust. Unless I've built it or am intimately familiar with it's quibbles and quirks, no way am I going to trust a scheduled reboot.

    Scheduled reboots are an important means of gaining trust.



  • @scottalanmiller absolutely, but I'll watch a few in person first haha



  • @Dashrender said:

    Why should I do it more often than that? Linux boxes have been known to go years without rebooting.

    Because it provides for safety and reliability. Just because machines CAN go without rebooting doesn't make it a good idea. And reboots are not to make systems run better, they are to protect against the unability to reboot when things are critical. No amount of "systems can run for years without a reboot" matters because that's not related to why we reboot.

    http://www.smbitjournal.com/2011/02/why-we-reboot-servers/