Hi all.
Im using a system as a basic multi purposes server, it had Fedora 26 with no issues. I tend to install linux, work on it installing this that and the other over time, forgetting half of the things ive done and half finished. So I like to start over after some time.
I clean installed Fedora 27 onto it, "dnf update"ed it and it installed kernel 4.14. 4.14 failed to work for me so ive stuck with 4.13 the OS came with. I have since installed F27 again but the problem persists.
Now I have installed 4.13 that works, 4.14 and 4.16 that fail. But I cant install VMware etc with newer kernels installed.
The issue with all these later kernels is they boot to console login, I enter my user name and press enter and it packs up. I get…
ata1.00: exception Emask 0x0 SAct 0x1ffc8000 SErr 0x0 action 8x6 frozen
failed command: READ FPDMA QUEUED
Loops over.
Some say its a drive error, but its a fresh install and still works with kernel 4.13? Any ideas? Holla what info you may need if helpful.
Thanks, TT24
UPDATE
See post 5 (with pic).
UPDATE
Just installed Fedora 27 clean again on another spare drive laying around and the same problem. 4.13 after OS install works fine. dnf update installs 4.16 and the second boot into it, I enter my username and press enter and it freezes. After some time I get errors ending up with…
Failed to open /var/log/journel
Do these things happen to everyone or is it system to system? I will continue researching.
Just for kicks, please try fedora 28.
Fedora 28 ships with kernel 4.16 by default so you should have an easier time.
2 Likes
Thanks very much for your reply.
I came home tonight with my time set on re installing and setting up this server as Ive pulled out a Mac I need to get data off of. Checked out Fedora’s site to see if their ISO is any newer and 28 1.1 is out. Will give it a go, sure it will work out fine. Will post back
1 Like
Damn and blast
Perhaps its kernel 4.16? But post install login kicked out the same errors as last time. Here’s a screenshot…
I could go back to Fedora 27 and block dnf from updating the kernel, but will likely cause problems, VM programs on the top of my head. I could use an earlier release but the newer kernel will probably install. I could stick with the 27 install I got working, but with the custom headers is a bit of a mess.
Stupid question but could this kernel issue be hardware related? Ill be working on this tonight but if anyone has any ideas please holla.
Thanks, TT24
Could you do me a favour and try booting with libata.force=noncq
?
You can enter it in grub by selecting the kernel in the bootloader, pressing ‘e’, going down to where the kernel options are, adding that to the line, and when you’re done pressing ctrl-x.
If that fixes your issue, you can make it permanent by adding it to the GRUB_CMDLINE_LINUX line in /etc/default/grub and running sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg
. This is assuming you’re running your system in UEFI mode, if not your grub.cfg will be located in the /boot directory, iirc.
Edit: what this does is disable NCQ.
Edit 2: Yes, this could be a hardware issue, could be a bad SATA cable, problems with the drive, or a problem with the SATA controller. Disabling NCQ might get the system running at least, but you should check to see if changing or simply reseating the cable fixes it. Also, you might want to check the drive with smartctl
.
1 Like
Thanks @magicthighs.
Sadly I dont think it worked. Too confirm I typed this and pressed ctl-x to boot…
But she still kicks off with this…
Ive removed all SATA cables, I tried another drive earlier. Im now using a new mSATA driver alone.
Im wondering if my best bet is to use Fedora 27 with kernel 4.13 and just stick to it? Blocking kernel updates from dnf shouldn’t cause issues? Im about to try Ubuntu 18.8 to see if that has kernel issues, then maybe Fedora Workstation.
With the kernel hardware question, I meant chipsets and the like. But cables are a good call.
Many thanks man, TT24.
EDIT
Ive just noticed the errors have chanced? But still related.
It’s odd that these errors exhibit when running one kernel but not the other. That could indicate it’s the SATA controller driver causing the problems. But yeah, if I were you I’d just stick with the working 4.13 kernel.
I’d still keep a very close eye on the drive, I’ve had drives with these issues in the past, and they ended up reallocating sectors soon after I first saw the errors. If you can get the system up with the 4.13 kernel you might want to back up the data on it just in case that starts happening to you.
And yes, the errors changed, before it was an error that usually indicates a problem with NCQ. Once you disable that the drives fall back to using DMA, which is what’s generating the errors now.
Thanks again man!
It is odd, yet another drive on another port now with Ubuntu 18.04 and shes doing the same. GUI login screen comes up, login and moments later it freezes then goes back to command prompt spewing out failed read, write etc.
Seems to happen after a short delay? Enter username fast enough and it will prompt for password then mess up, or like ubuntu before letting me enter credentials and making it to the desktop but then errors?
Anything after kernel 4.13. Will have a little more of a play, but I think sticking to 4.13 is my best option here.
Thanks again! TT24