Reboot your servers

JaredBusch

Always baffles me when you see a server hasn't been rebooted in over a year. It's my opinion that servers (all servers, regardless of the OS), should be rebooted as part of the monthly patch cycle, to test the hardware.

How could anyone know it will come back following a power outage, or something which forces a reboot...

@scottalanmiller and myself have posted more than once on this subject.

I reboot everything monthly.

scottalanmiller

That's why I wrote Why We Reboot Servers many years ago

scottalanmiller

I actually reboot weekly in most cases. Mostly this is because I find getting a reboot schedule is simpler on a weekly basis than a monthly. Like "Friday at 6PM" is reliable whereas "The First Day of the Month at Noon" will vary by day all over the place. Work tends to be more "day of the week" oriented than "day of the month." Unless it is a payroll processing system.

JaredBusch

@scottalanmiller I reboot on the weekend a week after patch Tuesdays. Makes life easy. I apply all the Windows updates, and reboot those, run any Linux updates, shut those down, then update the hypervisors and reboot those.

scottalanmiller

I directly only deal with the UNIX side. I schedule those to update themselves and then reboot when done updating; or we do auto updates during the week and then just reboot on the schedule.

wirestyle22

@scottalanmiller said in Reboot your servers:

I directly only deal with the UNIX side. I schedule those to update themselves and then reboot when done updating; or we do auto updates during the week and then just reboot on the schedule.

Zabbix monitoring is super convenient for my weekly reboots. If something doesn't come back I will receive an e-mail but I also remote into it to verify everything is up and running too

scottalanmiller

@wirestyle22 said in Reboot your servers:

Zabbix monitoring is super convenient for my weekly reboots. If something doesn't come back I will receive an e-mail but I also remote into it to verify everything is up and running too

We use this...

for i in $(cat server_list); do echo $i $(ssh $i uptime); done

Produces a nice list of the uptimes for every system.

coliver

We reboot the second Wednesday after patch Tuesday at 3am.

gjacobse

we have a pair that are on a schedule to reboot,... one is a SAP server that has print issues if running for two or three weeks without reboot. It's weird,.. but hey,... it's SAP.

StrongBad

@gjacobse said in Reboot your servers:

we have a pair that are on a schedule to reboot,... one is a SAP server that has print issues if running for two or three weeks without reboot. It's weird,.. but hey,... it's SAP.

Weird that it has managed to keep working for two weeks?

gjacobse

@StrongBad said in Reboot your servers:

@gjacobse said in Reboot your servers:

we have a pair that are on a schedule to reboot,... one is a SAP server that has print issues if running for two or three weeks without reboot. It's weird,.. but hey,... it's SAP.

Weird that it has managed to keep working for two weeks?

weird that printing quits..

Romo

@scottalanmiller said in Reboot your servers:

@wirestyle22 said in Reboot your servers:

Zabbix monitoring is super convenient for my weekly reboots. If something doesn't come back I will receive an e-mail but I also remote into it to verify everything is up and running too

We use this...
for i in $(cat server_list); do echo $i $(ssh $i uptime); done
Produces a nice list of the uptimes for every system.

I now do:

ansible all -a "uptime"

Texkonc

@gjacobse said in Reboot your servers:

we have a pair that are on a schedule to reboot,... one is a SAP server that has print issues if running for two or three weeks without reboot. It's weird,.. but hey,... it's SAP.

Hey my brother in law sells their stuff.

Texkonc

At work: every Friday for Workstations, if it is needed, and Every Sunday for Servers if it is needed.
Home: Autoreboot after they are installed on my laptop, cause I dont care, most my crap is web based. Home Lab: Every Sunday at 11AM.
My 1 rule, don't suck at patching.

scottalanmiller

@Texkonc said in Reboot your servers:

At work: every Friday for Workstations, if it is needed, and Every Sunday for Servers if it is needed.
Home: Autoreboot after they are installed on my laptop, cause I dont care, most my crap is web based. Home Lab: Every Sunday at 11AM.
My 1 rule, don't suck at patching.

What does "if it is needed" mean?

DustinB3403

I'm updating and rebooting my servers right now.

Texkonc

@scottalanmiller said in Reboot your servers:

@Texkonc said in Reboot your servers:

At work: every Friday for Workstations, if it is needed, and Every Sunday for Servers if it is needed.
Home: Autoreboot after they are installed on my laptop, cause I dont care, most my crap is web based. Home Lab: Every Sunday at 11AM.
My 1 rule, don't suck at patching.

What does "if it is needed" mean?

When patches are released, because you very well know, patching is the last thing customers think about. Although this is one thing I would like to audit as one of my other projects. New guy, fresh set of eyes.

scottalanmiller

@Texkonc said in Reboot your servers:

@scottalanmiller said in Reboot your servers:

@Texkonc said in Reboot your servers:

At work: every Friday for Workstations, if it is needed, and Every Sunday for Servers if it is needed.
Home: Autoreboot after they are installed on my laptop, cause I dont care, most my crap is web based. Home Lab: Every Sunday at 11AM.
My 1 rule, don't suck at patching.

What does "if it is needed" mean?

When patches are released, because you very well know, patching is the last thing customers think about. Although this is one thing I would like to audit as one of my other projects. New guy, fresh set of eyes.

Why not call "updating for updating's sake" "needed" and then update every week

gjacobse

This post is deleted!