3 Ubuntu distros go from fast reboot to SLOW after installing nVidia drivers

[finny388]

Intel(R) Core(TM) i5-4590 CPU @ 3.30GHz
8GB ram
gtx 970

boot-repair info:
http://paste.ubuntu.com/23223011/
(http://paste.ubuntu.com/23223011/)

Ubuntu Mate 16.04, Kubuntu 16.04, KDE Neon 5.7 (16.04)

Fresh installs and in all 3 cases they went from ~15 sec reboots to 2:30 to over 3 min after installing nVidia drivers. I'm currently on Neon and the driver is 361.42.


$ systemd-analyze blame
288ms dev-sdc2.device
283ms plymouth-start.service
192ms setvtrgb.service
155ms systemd-fsck@dev-disk-by\x2duuid-75d98f5c\x2d71f9\x2d4a25\x2dab69\x2d219e516f4e0b.s ervice
150ms upower.service
135ms networking.service
108ms ModemManager.service
89ms apport.service
88ms systemd-logind.service
84ms accounts-daemon.service
80ms apparmor.service
80ms thermald.service
69ms NetworkManager.service
69ms ondemand.service
66ms systemd-fsck@dev-disk-by\x2duuid-A4E0\x2d0DEF.service
65ms grub-common.service
63ms snapd.firstboot.service
56ms console-setup.service
50ms irqbalance.service
49ms plymouth-read-write.service
48ms systemd-udev-trigger.service
47ms gpu-manager.service
37ms mnt-FH_WD_BLACK_FILES.mount
35ms iio-sensor-proxy.service
35ms keyboard-setup.service
30ms systemd-journald.service
28ms packagekit.service
26ms udisks2.service
25ms systemd-udevd.service
24ms systemd-user-sessions.service
22ms systemd-tmpfiles-setup.service
22ms plymouth-quit.service

(there's more but just more milliseconds)

Any help or insights so appreciated.
Thanks

[QIII]

Hello!

If you are running Neon currently, this thread should be moved to the Ubuntu/Debian BASED forum.

If you would prefer that I don't move it from the Ubuntu Official Flavours Support area, please reboot to an official Ubuntu flavor and re-post the results given above.

[oldfred]

Do not know Neon.

Did you have the Neon drive as sda when you installed it?
With Ubuntu grub in UEFI mode only installs to drive seen as sda. But your ESP is sdc5.

Errors from systemd seem to be related to fsck on sdc2 and plymouth start?
Do not know plymouth issues.

But you may want to run a full fsck on sdc5 and perhaps other ext4 partitions.
#From liveDVD/Flash so everything is unmounted,swap off if necessary, change example shown with partition sdb1 to your partition(s)
#e2fsck is used to check the ext2/ext3/ext4 family of file systems. -p trys fixes where response not required
sudo e2fsck -C0 -p -f -v /dev/sdb1
#if errors: -y auto answers yes for fixes needing response, also see man e2fsck
sudo e2fsck -f -y -v /dev/sdb1

[finny388]

Okay, I'm logged into Kubuntu 16.04 with nVidia driver 364.19

Neon was always sdc. sdc is the main while sda is alternate (used to be main and pre-existed)
ESP?

UEFI confuses me.

Here's the systemd blame:

$ systemd-analyze blame
37.713s apt-daily.service
376ms accounts-daemon.service
345ms networking.service
321ms dev-sdc1.device
319ms thermald.service
316ms bluetooth.service
311ms avahi-daemon.service
307ms systemd-logind.service
307ms NetworkManager.service
301ms irqbalance.service
298ms grub-common.service
292ms iio-sensor-proxy.service
200ms wpa_supplicant.service
151ms upower.service
138ms systemd-fsck@dev-disk-by\x2duuid-A4E0\x2d0DEF.service
63ms systemd-rfkill.service
58ms apparmor.service
55ms snapd.firstboot.service
54ms console-setup.service
46ms systemd-udev-trigger.service
42ms apport.service
37ms ondemand.service
36ms systemd-journald.service
34ms keyboard-setup.service
30ms rsyslog.service
30ms systemd-modules-load.service
29ms systemd-udevd.service
26ms udisks2.service

[finny388]

Is there a tutorial for reading boot-info?

Seems there are instructions everywhere on how to create it and it ends there.

[oldfred]

Almost all of the Boot-Repair summary report is many standard terminal command's output.
Experience helps. Especially when it is something missing that should be there.
And good eyes, I still miss things that otherwise should be obvious.

I just ran this from my main working install of 16.04 on SSD:
fred@Asusz97:~$ systemd-analyze blame

8.936s NetworkManager-wait-online.service
3.771s apt-daily.service
520ms snapd.refresh.service
328ms systemd-fsck@dev-disk-by\x2duuid-a0c1c99f\x2d0f09\x2d4787\x2da7

And noticed my /mnt/data partition seemed to have issues. Rebooted into my Yakkety install which is on HDD and it loaded a lot slower. So ran fsck and this was new result on reboot. Only top output shown.

fred@Asusz97:~$ systemd-analyze blame

8.727s NetworkManager-wait-online.service
271ms dev-sda6.device
266ms dev-disk-by\x2duuid-493a8696\x2d5fa7\x2d48d7\x2d9ebb\x2d4c9ef24
146ms ModemManager.service
136ms systemd-fsck@dev-disk-by\x2duuid-a0c1c99f\x2d0f09\x2d4787\x2da7

[finny388]

Here's a scrape of the headings in boot-info:

Boot Info Summary:
Drive/Partition Info:
"ls -l /dev/disk/by-id" output:
Mount points:

sda1/boot/grub/grub.cfg:
sda1/etc/fstab:

sda8/boot/grub/grub.cfg:
sda8/etc/fstab:

sdc1/boot/grub/grub.cfg:
sdc1/etc/fstab:

sdc2/boot/grub/grub.cfg:
sdc2/etc/fstab:

sdd1/boot/grub/grub.cfg:
sdd1/syslinux.cfg:
sdd1: Version of COM32(R) files used by Syslinux:

StdErr Messages:
log of boot-repair 2016-09-24__03h37
os-prober:
blkid:

sdc1/etc/grub.d/ :
sdc1/etc/default/grub :
sdc2/etc/grub.d/ :
sdc2/etc/default/grub :
sda1/etc/grub.d/ :
sda1/etc/grub.d/40_custom :
sda1/etc/default/grub :
sda8/etc/grub.d/ :
sda8/etc/default/grub :

UEFI/Legacy mode:

PARTITIONS & DISKS:
parted -l:
parted -lm:
mount:
ls:
hexdump -n512 -C /dev/sdc4
hexdump -n512 -C /dev/sdc5
hexdump -n512 -C /dev/sdc6
df -Th:
fdisk -l:

Recommended repair

[oldfred]

So is there a question there?
Boot-Repair shows most of the commands, if you run them you need sudo for most. fdisk, parted, df, hexdump, blkid, etc all are commands, if you want more info and what parameters are use man pages
man parted
man df
i.e.
sudo fdisk -lu
sudo parted -l


Did you run the e2fsck and see if that made a difference?

[finny388]

I'll try es2fsck.

Stay tuned...

[finny388]

Running from the boot repair iso, command failed:

@lubuntu:~$ sudo e2fsck -C0 -p -f -v /dev/sdc5
e2fsck: Bad magic number in super-block while trying to open /dev/sdc5
/dev/sdc5:
The superblock could not be read or does not describe a valid ext2/ext3/ext4
filesystem. If the device is valid and it really contains an ext2/ext3/ext4
filesystem (and not swap or ufs or something else), then the superblock
is corrupt, and you might try running e2fsck with an alternate superblock:
e2fsck -b 8193 <device>
or
e2fsck -b 32768 <device>

[oldfred]

If sdc5 and the ESP, that is FAT32 and you have to use dosfsck.

dosfstools - dosfsck (aka fsck.msdos and fsck.vfat) utilities
Must be unmounted
sudo dosfsck -t -a -w /dev/sdc5
The -a seems to help in clearing dirty bit
https://bbs.archlinux.org/viewtopic.php?id=164185

[finny388]

seemed to run okay with no errors.
no noticeable change though

return code:
fsck.fat 3.0.26 (2014-03-07)
/dev/scd5: 371 files, 55188/98304 clusters

[oldfred]

Did you use the -a parameter in dosfsck to clear the dirty bit?

[finny388]

yes
sudo dosfsck -t -a -w /dev/sdc5

[finny388]

Switched back to nouveau driver and now it's fast again.
8 s for desktop to close, 6 sec more to POST, then about 30 s to get to desktop.

Old numbers were about 1:45 to POST and another 1 minute to get to desktop.


but I want to use nvidia, (nvidia works fine in Kubuntu 14.10)