LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   CentOS (https://www.linuxquestions.org/questions/centos-111/)
-   -   CentOS 7 disappointing (https://www.linuxquestions.org/questions/centos-111/centos-7-disappointing-4175515002/)

suicidaleggroll 08-15-2014 06:03 PM

CentOS 7 disappointing
 
Is anyone else as disappointed with CentOS 7 as I am?

The installer is an absolute abomination. I agree 100% with this review:
http://www.dedoimedo.com/computers/centos-7.html

But not only that, I couldn't even get it installed on my machine. I have no trouble with CentOS 3, 4, 5, 6, Debian 4, 5, 6, 7, Slackware, manual partitioning, etc. That's all a walk in the park, but this CentOS 7 installer (especially the manual partition configuration) is terrible.

Pick the disk, it says there's 120 GB available, click on manual partitioning, where is "Next"? Hunt all over the screen for "Next" or something like it. Oh, there's a tiny "Done" button in the very top left, why on Earth is it there and why does it say "Done", I'm not done. Click it anyway, it takes me to the manual configuration window. Click + to add a partition..."No space available on selected devices". What? You just said there was 120 GB available and it's a brand new blank drive that nothing has ever been written to. Click + again, same error. Screw it, let's click "-" and see what happens, nothing happens. Click "+" now? Oh now you're happy. Tell it the mount point, click Ok. Oops, didn't mean to pick ext3, can I edit it? Nope, apparently I have to delete it and start from scratch. Click "-"..."critical python error, system is rebooting". WTF?

Round 2 - try it again. Make sure you don't make ANY MISTAKES this time, since apparently you can't undo anything without seg faulting the installer. Go through all the menus just fine, start the installation. Installation finishes, system reboots, nothing happens, just an endless blinking cursor.

Round 3 - try it again. I don't remember seeing a bootloader option, let's look closer. Oh there it is, buried in TINY blue text at the very bottom of the screen...a bootloader is not being installed. Why not? Let's change that. Alright back to manual partitioning. There are the partitions I created from the last installation, let's just re-use them. Click on the partition. What? I can't assign it a mount point? Why the hell not? Click "-" to wipe out the root partition. Click "-" to wipe out the swap partition. "Critical python error, system is rebooting"

Screw it, let's just install CentOS 6...15 minutes later it's up and running without a problem.

I finally did get CentOS 7 installed in VirtualBox just to try it out. Gnome 3, yuck. The whole thing is slow, unresponsive. I wanted to do a yum update, but PackageKit locked out yum for TWENTY MINUTES! Why? It only takes a few seconds to refresh the repositories.

After an hour or so of screwing around with it, I just deleted the VM.

5 is great, 6 is great...what were they smoking when they put 7 together?

mihaua 08-15-2014 06:34 PM

CentOS RIP, same as other sheet - Fedora, RHEL, OpenMandriva, SUSE.
also they try to drown Debian and Arch with troyan horse with the name "systemd"

vermaden 08-18-2014 12:27 AM

Quote:

Originally Posted by suicidaleggroll (Post 5221799)
Is anyone else as disappointed with CentOS 7 as I am?

Lots of us.

What did You expected after systemd/journald merge? ;)

ReaperX7 08-18-2014 12:34 AM

I doubt it's just systemd, it could just be CentOS 7 was not properly finalized and tested with packages being poorly built, improperly patched, or it just doesn't work due to a severe bug somewhere.

I would honestly do a trace back through the system and look for the cause of the problem. What partitioning method did you use? What amount of swap did you set? Any underlying hardware problems?

Is there any other way to install CentOS using things like chroot and manual unpacking, editing, and other advanced setup techniques?

syg00 08-18-2014 01:21 AM

Hmmm - been a while since I looked at CentOS, so I tossed it on an old laptop a week or so back. Wasn't that bad, although the partitioner is lame - seriously lame.
Me, and most of the world, has come to terms with systemd so I didn't even notice it. But seeing as how all this bitching and moaning is going on I figured I'd re-install CentOS as a second system just to check the dual-boot/grub issues.
:doh: - I let it default the (non-/boot) lvs to XFS - and you can't shrink it like any sensible filesystem can.
Maybe I'm starting to agree with you all ... :D

Time for a re-install onto a full disk to check out that partitioner ...

ReaperX7 08-18-2014 01:36 AM

I never let a GUI tool partition my system. I use fdisk, cfdisk, gdisk, cgdisk, or parted.

syg00 08-18-2014 01:52 AM

Generally I agree - every distro I've ever used has had problems with the partitioner on the install. But this is a stand-alone test unit, and I specifically want to test the partitioner.

PrinceCruise 08-18-2014 05:15 AM

Count me in the disappointed list. I just wiped the CentOS partition and got trusty old Slack 14.1 back.

Regards.

suicidaleggroll 08-18-2014 01:41 PM

Quote:

Originally Posted by ReaperX7 (Post 5222769)
I would honestly do a trace back through the system and look for the cause of the problem. What partitioning method did you use? What amount of swap did you set? Any underlying hardware problems?

Manual partitioning, same as I do on all Linux installations. I allocated 2 GB swap, but with 8 GB of RAM it was kind of a formality since it shouldn't need anything given the machine's intended use. No hardware problems - brand new system, which is now working perfectly fine in Debian. Not cutting edge hardware either, your basic Supermicro workstation mobo, Intel Xeon E3-1245 CPU/GPU, 8GB ECC RAM, Intel enterprise SSD for the OS, and a pair of 4 TB platter drives for RAID 1 data storage (identified by the installer, but not used during the installation process, I set them up manually with mdadm afterwards).

Quote:

Originally Posted by ReaperX7 (Post 5222796)
I never let a GUI tool partition my system. I use fdisk, cfdisk, gdisk, cgdisk, or parted.

That's part of the problem. On attempt #3 I had the partitions all ready to go from attempt #2, but the installer wouldn't let me assign a mount point to them. On every Linux installer ever created, when a partition already exists on the disk you can either delete it, ignore it, or mount it somewhere, and if you choose to mount it you can also format it to a new FS type if you want.

The CentOS 7 installer will only let you delete and create, apparently. When you click on an existing partition, it tells you everything you could possibly want to know about it, but the "mount point" box is blank and greyed out. I spent several minutes combing through all of the options trying to find a way to just USE the existing partition for /, but I never could find it. So eventually I just deleted the partition so I could re-create it, and when I did that the installer errored out and the system rebooted.

It was by far the most frustrating installation experience I've ever had, including Windows 98, XP, Redhat (not RHEL) 7 back in ~2000, etc. It's like they jumped back two decades with regards to stability and usability.

Even something as simple as setting up NTP, a process that involves one button click in all other installers (Date and Time > Use NTP), is a convoluted mess here. The first option on the install menu is Date & Time, so you go there, there's even an option for using NTP, but wait! There's no network connection. So you have to exit and go back to the main screen, the very LAST option in the menu is Network, so you go there, then you have to turn on the network adapter (why was it not already on?), make sure it grabs an IP, then exit and go back to the first option, turn on NTP again, etc.

Here's a suggestion - how about if I tell you to use NTP and the network adapter is currently off, you, ya know, turn it on, instead of throwing an error like you don't know what's happening? That's what the Fedora 20 installer does. Or if you refuse to do that, how about making the network setup the first option, since it's clearly required before you can use NTP in Date & Time?

travisdh1 08-18-2014 03:04 PM

Yeah, the switch from 6 to 7 is painful. The installer (which has been used in Fedora since, what, 17?) is the worst one I can remember trying to figure out. I don't care that you can do everything all at the same time when it takes another 10 minutes for me to get everything "configured" like I want that's costing more time, all because I can't find where anything is..... one of my pet peeves, letting developers/engineers design user interfaces (seriously, I should never be doing UI!) Changing the boot managers, ok, I'll learn the new thing eventually, but I still can't see how this new thing is any better than what came before.

ReaperX7 08-18-2014 05:05 PM

If that's the case with the installer being that stupid, then CentOS 7 is a total bork. Flush it, and move on. It's not worth wasting time on a system that can't install or setup properly, and if the developers and publishers don't want to fix it, then find another distribution and let CentOS die.

It's happened to many distributions with piss-poor installers.

syg00 08-18-2014 09:21 PM

Anybody bother to read the (RHEL) Install Guide ?. No, me neither till later.
Remember this is built to be an opensource RHEL - they lay the ground-rules (more so now with the January announcement).

It's all documented as is Redhats want - even the highlighted warnings about filesystems being deleted unequivocally. I still reckon it's pretty lame in parts, but it is what it is. And in an enterprise setting where you just throw a bunch of luns at it, it probably works just fine.

travisdh1 08-19-2014 09:46 AM

Quote:

Originally Posted by syg00 (Post 5223434)
Anybody bother to read the (RHEL) Install Guide ?. No, me neither till later.
Remember this is built to be an opensource RHEL - they lay the ground-rules (more so now with the January announcement).

It's all documented as is Redhats want - even the highlighted warnings about filesystems being deleted unequivocally. I still reckon it's pretty lame in parts, but it is what it is. And in an enterprise setting where you just throw a bunch of luns at it, it probably works just fine.

That's the thing, this is RedHat, the fact that they would need to document the new installer should scream out "bad".

mr.b-o-b 08-19-2014 09:46 PM

The installer is a bit lame, but it's not that bad. I just loaded up a few new installs in VMware. I like how easy it is now to install the vmware tools :) I must admit I need to get used to systemd. Being a lifelong Slackware user by default I feel like I'm still getting used to SysV, and I must confess I grew quite fond of SysV. Only time will tell.

vermaden 08-20-2014 12:29 AM

This post greatly summarizes what happens in the Linux world since long time:
https://lkml.org/lkml/2014/8/12/459


All times are GMT -5. The time now is 01:49 AM.