New Admin - Questions about configuring and virtualizing existing servers on a new server.
Linux - NewbieThis Linux forum is for members that are new to Linux.
Just starting out and have a question?
If it is not in the man pages or the how-to's this is the place!
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
New Admin - Questions about configuring and virtualizing existing servers on a new server.
Background:
I am a new Red Hat Linux admin. My employer knew I had zero admin experience before I was hired, and they sent me to (the wrong) training (RapidTrack RHCE training).
I been maintaining the two existing servers we have here, updates, backups, etc... Nothing major, until this week.
Existing Servers - and issues:
Server 1: Dell PowerEdge 2950 with hardware issues (running RHEL 5)
Our small database server, which is a dual-core Xeon PowerEdge server (purchased in late 2006), crashed. The raid controller and motherboard are both failing, and it caused some corruption. It's back up and running, but the help I got from other network admins from my IT department looked at it with me, and they told me the hardware is on its way out, and that a new server should be purchased. This recommendation has been passed along up the chain of command, and a new server has been picked for me, and is in the process of being approved for purchase.
Server 2: Dell Optiplex (yes, a desktop machine) running RHEL 5.
This other server we have is used as a web server. Everything seems ok right now, but it's quite old, and isn't running on server hardware.
The New Server:
The new server will be a Dell PowerEdge R720, with a Xeon E5-2665 (8 core - 16 threads), 16GB 1600mhz memory, and (4) 1TB hard drives in Raid-5. (I can supply more specs if needed).
My plan with the New Server:
Since the new hardware is light years beyond what I am currently running on either server, I am thinking of virtualizing both existing servers, and running them on the new server. Otherwise, that's a lot of wasted hardware, that could be used better, and a good way to get my web server running on a more secure (in terms of hardware) environment.
My employer has a contract with VMWare, so I am able to get licenses and the software to run virtualized environments.
I also have contracts with Red Hat, and will be able to get RHEL6, etc...
Opinions and questions:
I am likely going to be completely on my own with the project with no help here.
As a novice, am I being overly ambitious? Is what I want to do even possible, or am I biting off more than a novice can chew?
Are there any suggested books out there for taking existing servers and virtualizing them?
If it was you, would you do two clean virtual builds with RHEL6, and build a new machine from your backups, or would you just grab current machines and virtualize them in their existing form?
Distribution: K/Ubuntu 18.04-14.04, Scientific Linux 6.3-6.4, Android-x86, Pretty much all distros at one point...
Posts: 1,802
Rep:
Get Mike Jang's RHCSA/RHCE book...
The preferred RedHat way to virtualize is KVM,... So, you'll have to adapt your techniques to VMware, should you choose to go that route... Note that if you ever discontinue those licenses (RH & VMware), while using VMware, you'll have a migration problem. If you go with KVM, you'll be making a decision to do things differently than your company's SOP,... so clear it first,... But if you go with KVM, you'll be able to migrate the server to CentOS or Scientific Linux in the future with fewer issues.
You should be able to migrate the servers, but it'll be tough to do so without interruption or a lot of work... What you need to do is be able to backup everything you have and then install new virtual servers, running whatever you decide.
Then you need to migrate the services, not duplicate them. I advise against just trying to image drives and dump those images on a VM. What you need to do is try using the recommended backup and restore procedures for the running services, and restore to the new virtual machines. Use the existing machines configurations as a guide as how to configure the new VMs. Audit them as a first step in determining how they are set up,... the inits, configured services, installed packages, ... everything. Look at the directory structure, users, groups,... the whole shebang...
Then double check to make sure the virtual servers are running correctly, without issues, before going live with them. You'll need to transition them at off-peak hours. For a while you should attempt to keep the old machines ready to put back in place if there are problems. Oh, and don't forget to clone your VMs once they are up and running, keep the clones in a safe, accessible storage location.
I have also encountered this error, but only a few times. The problem was due to mismatch inn vmware converter and esx versions. Give it a try and post again if it fails.
I told the directors my plans, and they agreed virtualizing both of my current servers would be a great idea. They were a little concerned about memory needs, so they are going back to the vendor asking them to prepare another quote with 32GB of ram, but they are thinking of removing one of the drives from the configuration.
Question about RAID-5: If I recall back, RAID-5 with four drives will continue to run with only 3 working drives, but with only 3 drives, it will stop running to prevent any data loss. Is this correct?
I could easily get by one four 500GB Drives in RAID 5, as my current servers are about 12GB and 30GB in total size. The largest I heard is about 70-75GB for my web server, because of the video files we need to have on it for students.
Distribution: K/Ubuntu 18.04-14.04, Scientific Linux 6.3-6.4, Android-x86, Pretty much all distros at one point...
Posts: 1,802
Rep:
You need to do a cost/benefit analysis. I'd recommend seriously looking at dropping the size of the drives and seeing if you can get 4 or 5 smaller drives (versus 3) for a Raid 6. It doesn't seem like your workplace does intensive writes to disk, so that takes care of the potential write penalty. Raid 6 can (theoretically) lose up to two disks and still allow for a rebuild. If you can get them to spring for five 500GB drives, you'd have Raid 6's minimum 4, plus a hot spare. You might be able capitalize on any raw nerves they may have had regarding the rebuilding of the last failed Raid...
32 GB RAM seems like a good idea, especially for potential future expansion. The processor seems like it is more than enough to handle what you currently must handle several times over. CPUs can be overbooked in virtualization. RAM can't.
If you are on your own VMware is a good way to go as administration can be quick and easy (usually) compared to KVM. Not trying to trigger a VMware / KVM war, I use both, this is just my opinion for my situation. Get as much ram as you can and the fastest disks possible.
RAID 5 will tolerate one disk failure. The PERC controllers usually let you run a hot spare as well. If this is your only basket to put the eggs in I would do RAID 5 plus hot spare or RAID 6, 1+0, 0+1 depending on what is important and far as read write performance. The slicing of the array luns isn't a huge deal with the later ESXi versions, there used to be some LUN locking that needed to be planned if VMs were going to be changing states much.
As far as disk usage you could always start by thin provisioning the VM disks.
CPUs can be overbooked in virtualization. RAM can't.
I was always under the impression that you could over commit on RAM just fine esp when trimming is enabled. In fact if you have similar VMs there should be some page sharing benefits as well (TPS).
Of course pushing it to the max could create a performance issue. So the premise of not doing this is certainly valid.
Distribution: K/Ubuntu 18.04-14.04, Scientific Linux 6.3-6.4, Android-x86, Pretty much all distros at one point...
Posts: 1,802
Rep:
Quote:
Originally Posted by davx
I was always under the impression that you could over commit on RAM just fine esp when trimming is enabled. In fact if you have similar VMs there should be some page sharing benefits as well (TPS).
Of course pushing it to the max could create a performance issue. So the premise of not doing this is certainly valid.
I'm not a fan of VMware, mostly due to license costs and being strictly proprietary, but looking into it more, apparently, it handles memory overbooking fairly well, comparatively speaking,... My mindset is KVM right now...
But as you said, it's a performance issue. In this case, I think he's got the room. There are only 2 virtual servers he'll be running (headless, I assume),... He's got less file space, between the two, being utilized than I have on my home server... I can only assume that resources won't be in a pinch...
If you are on your own VMware is a good way to go as administration can be quick and easy (usually) compared to KVM. Not trying to trigger a VMware / KVM war, I use both, this is just my opinion for my situation. Get as much ram as you can and the fastest disks possible.
RAID 5 will tolerate one disk failure. The PERC controllers usually let you run a hot spare as well. If this is your only basket to put the eggs in I would do RAID 5 plus hot spare or RAID 6, 1+0, 0+1 depending on what is important and far as read write performance. The slicing of the array luns isn't a huge deal with the later ESXi versions, there used to be some LUN locking that needed to be planned if VMs were going to be changing states much.
As far as disk usage you could always start by thin provisioning the VM disks.
I will have RAID 5 and I believe a hot spare. The hardware is out of my hands, I don't have much of a choice with it. I'm just happy that it appears to be extreme overkill for what I want to do.
I was learning towards VMWare because this is something we already have licensed here with contracts, and I am looking for quick and easy. In addition to being a very novice server admin, I am also tasked with web development, database management, and managing two labs. (I am pretty damn good with those things, but they sap time away from me learning how to be a good server admin)
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.