Upgrade from Legacy or not?

#1
Been with KH for quite some time and am on a legacy VPS3 w/3.5 MB RAM. My load avg. so far is 0.2 :)

Now I'll be hosting about 1/2 dozen WordPress sites and am wondering if it's worthwhile upgrading to the Cloud-2. My primary concern is the server crashing and as I understand it Cloud hosting is more reliable. Y/N?

Also, I just experienced an issue where my connection speed was 100Mb! It took 5 hours to download a WP backup. Support increased the network speed to 1000Mb and you can imagine the result. Less than 10 minutes. So I'm wondering what connection speeds are the default and why mine were so low.
 

KH-Jonathan

Director of Managed Services
Staff member
#2
Been with KH for quite some time and am on a legacy VPS3 w/3.5 MB RAM. My load avg. so far is 0.2 :)

Now I'll be hosting about 1/2 dozen WordPress sites and am wondering if it's worthwhile upgrading to the Cloud-2. My primary concern is the server crashing and as I understand it Cloud hosting is more reliable. Y/N?

Also, I just experienced an issue where my connection speed was 100Mb! It took 5 hours to download a WP backup. Support increased the network speed to 1000Mb and you can imagine the result. Less than 10 minutes. So I'm wondering what connection speeds are the default and why mine were so low.
Going to MVPS-* line or the Cloud VPS line would definitely be faster for you as they're both run on SSDs instead of the traditional spinning disks your VPS-3 is on.

How big was the backup you were downloading? 100Mbps for 5 hours is 225GB; quite significant. We don't offer any VPSs with port speeds over 100Mbps and 100Mbps became the default several years ago (from 10Mbps - though when it was 10 we'd raise it to 100 for people). Being somewhat familiar with that ticket the speed was adjusted back to 100 as it never should've been set to 1000.
 
#3
Well Jonathan, something happened last nite with KH because as I said, that open ftp server gave me the expected download speeds. You can check the screenshot I included with my support request to see what KH was.

So, reading the KH site it seems, and correct me if I'm wrong, that if one cloud instance goes down, the sites on it also go down. So it's really not like a raid box where there's a hot swap spare? In which case I don't see a reason to pay for cloud if it's reliability is no different than an MVPS. Reason I'm asking is that the Forum "Network and Hardware status" currently shows a cloud failure. As well, I've been monitoring that forum and altho there's no history, it seems the SSD's are suffering from 'panic' failures more so than the legacy. Down time is my main concern. Speed of SSD vs. legacy, not so much.
 

KH-Jonathan

Director of Managed Services
Staff member
#4
I'll check the ticket.

If the physical piece of hardware (server, we call them nodes) goes down yes the VPSs on it go down. That's how it works regardless of traditional VPS, cloud, etc. A VM can't run in thin air. It's the same on Amazon's cloud or Microsoft's, etc.

The difference comes into play on how quickly we can restore service after such an outage. In this case it was 2-3 minutes and we had people back online. Had this been an MVPS node even with SSDs that would be at least 10-15 minutes, and had there been a hardware fault it would've potentially taken an hour or two depending on the situation. With Cloud we can boot people on other hardware instantly, even if the whole hardware node caught fire and was a melted puddle of metal. Ok that's a bit drastic, but you get the point.

on MVPS-* nodes (OpenVZ) we've identified most of the crashes being cause by KernelCare (which our Cloud nodes also run, though on a different kernel version) so we disabled KernelCare on MVPS-* nodes and haven't had any crashes since. This particular cloud node that crashed wasn't running KernelCare so it wasn't the culprit.

Sometimes freak kernel panics occur, sometimes there's more to the story. Since it just happened we don't yet know the culprit as our first focus is always getting everyone back online.
 
#5
Thanks for the clarification.

The other day when I logged into WHM it said to reboot so that KernelCare could be installed. Is my VPS at risk?

And oh, is the RAM on the new servers dedicated or burst?
 

KH-Jonathan

Director of Managed Services
Staff member
#6
Thanks for the clarification.

The other day when I logged into WHM it said to reboot so that KernelCare could be installed. Is my VPS at risk?

And oh, is the RAM on the new servers dedicated or burst?
On OpenVZ VPSs you don't have to worry about the kernel and we do use KernelCare. cPanel erroneously reports on it a lot but you need not worry about it.

We'll return kernelcare to the MVPS-* servers after the issues are worked out on their end.

Yeah, we identified a stick of RAM that died a pretty tragic death while reviewing logs. Started digging deeper and realized the system wasn't even seeing it at all so it's really dead. This particular hardware node is only about 5 months old at this point. Just an unfortunate system that nothing could've really prevented, sometimes catastrophic hardware failure just happens.

Goes to show that if you take away (read: the stick dies) 64GB of memory from a system at random it does not like it and crashes :p
 

KH-Derrick B.

Support Manager
Staff member
#7
The KernelCare notice on OpenVZ servers is actually a cPanel bug -- it's been reported as fixed in WHM v78.0.3, but I'd recommend waiting until v78 hits stable before upgrading.
 
Top