ESX advise

[adam1942]

me and my mate where discussing our ESX hosts and he said that he was going to have to reboot his because they where up for around a year.. I was telling him that it shouldnt really matter and we had had ours up for over about a year and a half so far.. no need to reboot unless theres a problem as its running a modified *nix kernal right? whats everyones take on this? cheers

 

[Mike_182]

Hosts should only need restarting when they need it. Even my Windows stuff stays up for 2-3 years on end.

 

[adam1942]

grand so thats what i thought hes just worried incase his esx host dies with the 20+ vms on it

we set it up though and theres enough room on each of the other two vms so it should fail over. cheers for the reply mike

 

[DMS]

Hosts should only need restarting when they need it. Even my Windows stuff stays up for 2-3 years on end.

+1.

A customer of ours has an old NT4 box which acts purely as a RADIUS server (using Steel Belted RADIUS server I believe) and nothing more. It's been up for well over 2 years without issue.

It's a pity half the Windows updates these days require the machines to be restarted otherwise patching them (and maintaining availability) wouldn't be such a ball ache.

 

[ChrisR]

+1.

A customer of ours has an old NT4 box which acts purely as a RADIUS server (using Steel Belted RADIUS server I believe) and nothing more. It's been up for well over 2 years without issue.

It's a pity half the Windows updates these days require the machines to be restarted otherwise patching them (and maintaining availability) wouldn't be such a ball ache.

I was about to say I'm sure there's been server updates in the last 2 years that have needed a reboot

 

[Longy]

Aye I've got a few SBS customers and I swear I have to reboot them weekly because of bloody updates.

 

[dk]

I was about to say I'm sure there's been server updates in the last 2 years that have needed a reboot

+1, theres no way a windows server stays up more than 6 months these days with updates, 3 years means you aren't patching and thats pretty bad practice imo!

 

[dk]

Aye I've got a few SBS customers and I swear I have to reboot them weekly because of bloody updates.

servers like the ocassional reboot, we must reboot ours every month or so either for windows updates or software updates etc.

We have a couple of servers that surprise us when we see they have been going for 6 months or so but they get a restart every now and again too.

with esx, i suppose they don't really need rebooting, but again, they also have updates to apply, especially version upgrades like the current version is esx 3.5 update 4, those updates require restarts normally and if you aren't applying them then you aren't following best practices.

 

[ChrisR]

There's nothing wrong with rebooting a server, unless you're an uptime stat slut

I'd say it's extremely rare at our place if a windows box hasn't been rebooted within a month or 2 at the most.

 

[DMS]

I agree, there's nothing wrong with rebooting servers. The main problem in a business environment is usually availability. Some of our customers pay for 99.9% uptime for some of their business critical services and applications. That's why we have to implement enormous CSF's with huge server clusters connected to highly redundant SAN's so that if a server fails or is rebooted, availability isn't affected. If we don't meet the availability that the customer requires, we get bent over and shafted when it comes to paying penalty fees.
One customer of ours has an entire /21 subnetted LAN (with only a few addresses spare) just for the heartbeat function.

 

[Mike_182]

The trouble is where you walk into an environment where you're picking up the pieces of someone else's lack of understanding - somewhere where the last technician thought High Availability was plugging in both power supplies to the same power rail - then you can't just go restarting kit...