Announcement

**oshunluvr** · Nov 25, 2014, 08:26 AM

I can't imagine why it doesn't go to UDMA6 - I would look at your mobo drive controller first, but I can say you should not have trim (discard) enabled in fstab. There's no reason to and it slows things down and causes excessive wear. Either run trim manually a couple times a year or set a cron job to do it no more than monthly.

Besides, the cached reads go through the system cache buffer. Your SSD is reading 3 times faster when read from directly so it's not like your SSD is slower than the HDDs. Although, I get why you want max performance.

EDIT: Also use reatime in your mount options for the SSD.

**bonkers** · Nov 25, 2014, 07:34 PM

This is my fstab could you please tell me anything I am missing? All this is on one SSD, any idea why the UUID is different for each partition, is this normal?

Code:

#
# Use 'blkid' to print the universally unique identifier for a
# device; this may be used with UUID= as a more robust way to name devices
# that works even if disks are added and removed. See fstab(5).
#
# <file system> <mount point>   <type>  <options>       <dump>  <pass>
# / was on /dev/sda1 during installation
UUID=c2590e88-813b-4481-b1e2-cfc449cedb00 /               ext4    noatime,nodiratime,commit=600,errors=remount-ro 0       1
# /home was on /dev/sda2 during installation
UUID=ac1aefd2-69aa-4ce2-9341-3f28a1c816b9 /home           ext4    defaults,noatime       0       2
# swap was on /dev/sda5 during installation
UUID=f7c78b56-7e43-40b7-8764-f1b69c256066 none            swap    sw              0       0

**SteveRiley** · Nov 25, 2014, 08:42 PM

I also have a Crucial M4. Here are my timing reads:

Code:

steve@t520:~$ [B]sudo hdparm -Tt /dev/sda[/B]

/dev/sda:
 Timing cached reads:   11982 MB in  2.00 seconds = 5994.08 MB/sec
 Timing buffered disk reads: 1084 MB in  3.00 seconds = 361.27 MB/sec

And info about the drive itself:

Code:

steve@t520:~$ [B]sudo hdparm -I /dev/sda[/B]

/dev/sda:

ATA device, with non-removable media
        Model Number:       M4-CT256M4SSD2                          
        Serial Number:      000000001140031E608A
        Firmware Revision:  0009    
        Transport:          Serial, ATA8-AST, SATA 1.0a, SATA II Extensions, SATA Rev 2.5, SATA Rev 2.6, SATA Rev 3.0
Standards:
        Used: unknown (minor revision code 0x0028) 
        Supported: 9 8 7 6 5 
        Likely used: 9
Configuration:
        Logical         max     current
        cylinders       16383   16383
        heads           16      16
        sectors/track   63      63
        --
        CHS current addressable sectors:   16514064
        LBA    user addressable sectors:  268435455
        LBA48  user addressable sectors:  500118192
        Logical  Sector size:                   512 bytes
        Physical Sector size:                   512 bytes
        Logical Sector-0 offset:                  0 bytes
        device size with M = 1024*1024:      244198 MBytes
        device size with M = 1000*1000:      256060 MBytes (256 GB)
        cache/buffer size  = unknown
        Form Factor: 2.5 inch
        Nominal Media Rotation Rate: Solid State Device
Capabilities:
        LBA, IORDY(can be disabled)
        Queue depth: 32
        Standby timer values: spec'd by Standard, with device specific minimum
        R/W multiple sector transfer: Max = 16  Current = 16
        Advanced power management level: 254
        DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 [B][COLOR="#B22222"]*udma5 [/COLOR][/B]
             Cycle time: min=120ns recommended=120ns
        PIO: pio0 pio1 pio2 pio3 pio4 
             Cycle time: no flow control=120ns  IORDY flow control=120ns

And yes, it's reporting as UDMA5. The thing to remember, though, is that UDMA interfaces are for the older parallel ATA standard. They have absolutely no releationship to modern Serial ATA classifications or specifications. hdparm is just reporting what it sees through your compter's firmware PATA compatibility layer.

In your fstab, nodiratime is redundant. noatime covers both files and directories. And yes, it's normal to see separate UUIDs for each filesystem. In fact, take a look inside /dev/disk -- you'll see several different naming structures.

**oshunluvr** · Nov 25, 2014, 09:05 PM

You can change noatime and nodirtime to relatime (or delete them) but the impact is very small (Linus approves if that matters to you

). relatime is now the default in Ubuntu.

Yes, different UUIDs is not only normal, it's required. Each file system needs to have a different UUID so various system parts know what is what. Universally Unique IDentifier...

thanks Steve!

**bonkers** · Nov 26, 2014, 09:07 AM

Everytime I bootup my computer my dmesg shows this first [ 2.282841] EXT4-fs (sda1): mounted filesystem with ordered data mode. Opts: (null)
Then a few lines later, this EXT4-fs (sda1): re-mounted. Opts: commit=600,errors=remount-ro

Why does it not mount it correctly the first time around? Any reason? Also, is there a way to check if noatime was started, would it show up on dmesg if it was?
I switched my fstab as you guys advised to this:

Code:

# /etc/fstab: static file system information.
#
# Use 'blkid' to print the universally unique identifier for a
# device; this may be used with UUID= as a more robust way to name devices
# that works even if disks are added and removed. See fstab(5).
#
# <file system> <mount point>   <type>  <options>       <dump>  <pass>
# / was on /dev/sda1 during installation
UUID=c2590e88-813b-4481-b1e2-cfc449cedb00 /               ext4    noatime,nodiratime,commit=600,errors=remount-ro 0       1
# /home was on /dev/sda2 during installation
UUID=ac1aefd2-69aa-4ce2-9341-3f28a1c816b9 /home           ext4    defaults,noatime       0       2
# swap was on /dev/sda5 during installation
UUID=f7c78b56-7e43-40b7-8764-f1b69c256066 none            swap    sw              0       0
tmpfs           /tmp tmpfs                                        defaults,noatime,mode=1777 0       0

I added the tmpfs line at the end, is this correct way of doing it? And will It help me any?

**oshunluvr** · Nov 26, 2014, 10:47 AM

Yes to the tmpfs except defaults? Why have default followed by a non-default? Just seems unnecessary. I can't say it's bad.

Mine is:

#TMPFS in RAM
tmpfs /tmp tmpfs noatime,mode=1777,size=16G 0 0

The size option tells tmpfs to report 16G even though I don't have that much RAM. It swaps out to the swap partition if I go over 16G. I think default is half RAM size.

To tell your current mount options cat /proc/mounts or mount should show them all. I can't say why the remounting occurs except maybe because it's your boot partition for grub and gets mounted by grub with different options? I boot to a different device so I can't check that at the moment. I also don't use EXT-4 much either so it might be related to that.

**dibl** · Nov 26, 2014, 12:33 PM

udma5 mode is not necessarily slow:

Code:

root@imerabox:/# hdparm -i /dev/sda

/dev/sda:

 Model=OCZ-REVODRIVE, FwRev=1.37, SerialNo=OCZ-Q17BKCW4VFG6ZBG7
 Config={ Fixed }
 RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
 BuffType=unknown, BuffSize=unknown, MaxMultSect=16, MultSect=16
 CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=117231408
 IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes:  pio0 pio1 pio2 pio3 pio4 
 DMA modes:  mdma0 mdma1 mdma2 
 UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6 
 AdvancedPM=no WriteCache=enabled
 Drive conforms to: unknown:  ATA/ATAPI-2,3,4,5,6,7

 * signifies the current active mode

root@imerabox:/# hdparm -Tt /dev/sda

/dev/sda:
 Timing cached reads:   31716 MB in  2.00 seconds = 15880.28 MB/sec
 Timing buffered disk reads: 566 MB in  3.00 seconds = 188.58 MB/sec

Here are a couple of tips for SSD installation/configuration:

- Use ext4, but use the "commit" mount option to slow down the journal-writing frequency. I use commit=120, a 24x reduction over the default 5 seconds.
- You can also use a tmpfs mount for /var/log, after you have confirmed that your installation is stable. You will lose the logs every reboot.
- You can configure firefox and also chrome to change the default cache location. For example, you can make a directory in /run/user/1000 (for single user system) for the cache, and then use the browser's configuration settings to have it do its caching there. All assuming you have plenty of memory on board.
- You can set a cron job to run fstrim weekly or monthly, and there's a great script to do that, which I will post here for convenience. It came from here:

Code:

#!/bin/sh
#
# To find which FS support trim, we check that DISC-MAX (discard max bytes)
# is great than zero. Check discard_max_bytes documentation at
# https://www.kernel.org/doc/Documentation/block/queue-sysfs.txt
#
for fs in $(lsblk -o MOUNTPOINT,DISC-MAX,FSTYPE | grep -E '^/.* [1-9]+.* ' | awk '{print $1}'); do
	fstrim "$fs"
done

and here is my script that goes in ~/.kde/Autostart to make the browser cache directories at login:

Code:

#!/bin/bash
NEWDIR=/run/user/1000/chromium-cache
mkdir -p "$NEWDIR" &
sleep 1
NEWDIR1=/run/user/1000/firefox-cache
mkdir -p "$NEWDIR1" &
#end

**SteveRiley** · Nov 26, 2014, 01:09 PM

Originally posted by bonkers View Post

Everytime I bootup my computer my dmesg shows this first [ 2.282841] EXT4-fs (sda1): mounted filesystem with ordered data mode. Opts: (null)
Then a few lines later, this EXT4-fs (sda1): re-mounted. Opts: commit=600,errors=remount-ro

Why does it not mount it correctly the first time around? Any reason?

Mount options can be stored in the superblock of an EXT2/EXT3/EXT4 file system. They can also be specified on the entry in fstab. The first line in dmesg is reporting what the kernel sees when it mounts the file system. The second line is reporting what happens after reading fstab. Here, because you have changed some of the defaults, the kernel needs to remount the file system. For reasons I've never quite understood, the noatime (and related) options don't show up in the log entry, even though they are honored. As Oshun mentioned, you can verify this by taking a look at the file /proc/mounts.

**bonkers** · Nov 26, 2014, 04:19 PM

Originally posted by dibl View Post

udma5 mode is not necessarily slow:

Code:

root@imerabox:/# hdparm -i /dev/sda

/dev/sda:

 Model=OCZ-REVODRIVE, FwRev=1.37, SerialNo=OCZ-Q17BKCW4VFG6ZBG7
 Config={ Fixed }
 RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
 BuffType=unknown, BuffSize=unknown, MaxMultSect=16, MultSect=16
 CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=117231408
 IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes:  pio0 pio1 pio2 pio3 pio4 
 DMA modes:  mdma0 mdma1 mdma2 
 UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5 udma6 
 AdvancedPM=no WriteCache=enabled
 Drive conforms to: unknown:  ATA/ATAPI-2,3,4,5,6,7

 * signifies the current active mode

root@imerabox:/# hdparm -Tt /dev/sda

/dev/sda:
 Timing cached reads:   31716 MB in  2.00 seconds = 15880.28 MB/sec
 Timing buffered disk reads: 566 MB in  3.00 seconds = 188.58 MB/sec

Here are a couple of tips for SSD installation/configuration:

- Use ext4, but use the "commit" mount option to slow down the journal-writing frequency. I use commit=120, a 24x reduction over the default 5 seconds.
- You can also use a tmpfs mount for /var/log, after you have confirmed that your installation is stable. You will lose the logs every reboot.
- You can configure firefox and also chrome to change the default cache location. For example, you can make a directory in /run/user/1000 (for single user system) for the cache, and then use the browser's configuration settings to have it do its caching there. All assuming you have plenty of memory on board.
- You can set a cron job to run fstrim weekly or monthly, and there's a great script to do that, which I will post here for convenience. It came from here:

Code:

#!/bin/sh
#
# To find which FS support trim, we check that DISC-MAX (discard max bytes)
# is great than zero. Check discard_max_bytes documentation at
# https://www.kernel.org/doc/Documentation/block/queue-sysfs.txt
#
for fs in $(lsblk -o MOUNTPOINT,DISC-MAX,FSTYPE | grep -E '^/.* [1-9]+.* ' | awk '{print $1}'); do
    fstrim "$fs"
done

and here is my script that goes in ~/.kde/Autostart to make the browser cache directories at login:

Code:

#!/bin/bash
NEWDIR=/run/user/1000/chromium-cache
mkdir -p "$NEWDIR" &
sleep 1
NEWDIR1=/run/user/1000/firefox-cache
mkdir -p "$NEWDIR1" &
#end

I am very interested in what you have written, do you mind sharing with me the fstab entry for firefox cache (by the way, I have 16gb of ram)and the fstab entry to put /var/log as tmpfs. Secondly, I checked the /run/user/1000 directory and I found some interesting things in there such as: /run/user/1000/upstart/sessions/ and inside this dir is a file named 1884.session which reads

Code:

UPSTART_SESSION=unix:abstract=/com/ubuntu/upstart-session/1000/1884

, why is there an 1884, is this normal? Also a .txt file called /run/user/1000/upstart-file-bridge.1884.pid with this inside it

Code:

here is a picture:

I am also not allowed to change the permissions for the root directory /run/user/1000 I can untick others, ownings group, but can't apply, this is pretty much the case for anything in root. This is what it looks like:

I want to thank you for your time, it means a great deal to me, all you guys are teaching me quite a bit in a rather fast pace. You know, I started out with ubuntu dapper drake back in 2006? was it, I believe it was 6.06LTS and I was even able to complie a kernel, albeit it seems it was much easier back then, I used the master kernel thread, I would grab the bunzip, extract to /usr/src, then cd into it, but I used xconfig as well, I was a noob then, but I was actually able to compile it and install it, along with the nvidia.sh file using init3 then init5, but I can't seem to be able to compile a kernel from source for 14.04, so much has changed and I can get as far as xconfig, but once it is configured, I get stuck, but that is another story, sorry for getting off track, but I stopped using *buntu after about a year (life got in the way), otherwise I would probably be much more efficient than I am now. I am very embarrased at some of the questions I ask, I am sure this is a *small beer* as my grandfather says to you guys, but for me, this is not easy at all. But thank you all VERY much, I mean it.

**bonkers** · Nov 26, 2014, 04:24 PM

Originally posted by SteveRiley View Post

Mount options can be stored in the superblock of an EXT2/EXT3/EXT4 file system. They can also be specified on the entry in fstab. The first line in dmesg is reporting what the kernel sees when it mounts the file system. The second line is reporting what happens after reading fstab. Here, because you have changed some of the defaults, the kernel needs to remount the file system. For reasons I've never quite understood, the noatime (and related) options don't show up in the log entry, even though they are honored. As Oshun mentioned, you can verify this by taking a look at the file /proc/mounts.

Does this look correct to you?

Code:

rootfs / rootfs rw 0 0
sysfs /sys sysfs rw,nosuid,nodev,noexec,relatime 0 0
proc /proc proc rw,nosuid,nodev,noexec,relatime 0 0
udev /dev devtmpfs rw,relatime,size=8064936k,nr_inodes=2016234,mode=755 0 0
devpts /dev/pts devpts rw,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000 0 0
tmpfs /run tmpfs rw,nosuid,noexec,relatime,size=1621624k,mode=755 0 0
/dev/disk/by-uuid/c2590e88-813b-4481-b1e2-cfc449cedb00 / ext4 rw,noatime,errors=remount-ro,commit=600,data=ordered 0 0
none /sys/fs/cgroup tmpfs rw,relatime,size=4k,mode=755 0 0
none /sys/fs/fuse/connections fusectl rw,relatime 0 0
none /sys/kernel/debug debugfs rw,relatime 0 0
none /sys/kernel/security securityfs rw,relatime 0 0
none /run/lock tmpfs rw,nosuid,nodev,noexec,relatime,size=5120k 0 0
none /run/shm tmpfs rw,nosuid,nodev,relatime 0 0
none /run/user tmpfs rw,nosuid,nodev,noexec,relatime,size=102400k,mode=755 0 0
none /sys/fs/pstore pstore rw,relatime 0 0
/dev/sda2 /home ext4 rw,noatime,data=ordered 0 0
/dev/sdi1 /media/movies1720 ext4 rw,relatime,data=ordered 0 0
/dev/sde1 /media/1080movies2 ext4 rw,relatime,data=ordered 0 0
/dev/sdg1 /media/Torrents ext4 rw,relatime,data=ordered 0 0
/dev/sdd2 /media/Shows2 ext4 rw,relatime,data=ordered 0 0
/dev/sdb1 /media/Movies2_720p ext4 rw,relatime,data=ordered 0 0
/dev/sdc1 /media/movies1080 ext4 rw,relatime,data=ordered 0 0
/dev/sdh1 /media/Series720and1080 ext4 rw,relatime,data=ordered 0 0
/dev/sdf1 /media/Shows ext4 rw,relatime,data=ordered 0 0
systemd /sys/fs/cgroup/systemd cgroup rw,nosuid,nodev,noexec,relatime,name=systemd 0 0

With this fstab file onboard:

Code:

# /etc/fstab: static file system information.
#
# Use 'blkid' to print the universally unique identifier for a
# device; this may be used with UUID= as a more robust way to name devices
# that works even if disks are added and removed. See fstab(5).
#
# <file system> <mount point>   <type>  <options>       <dump>  <pass>
# / was on /dev/sda1 during installation
UUID=c2590e88-813b-4481-b1e2-cfc449cedb00 /               ext4    noatime,commit=600,errors=remount-ro 0       1
# /home was on /dev/sda2 during installation
UUID=ac1aefd2-69aa-4ce2-9341-3f28a1c816b9 /home           ext4    defaults,noatime       0       2
# swap was on /dev/sda5 during installation
UUID=f7c78b56-7e43-40b7-8764-f1b69c256066 none            swap    sw              0       0
tmpfs           /tmp tmpfs                                        defaults,noatime,mode=1777 0       0
/dev/sdb1       /media/Movies2_720p       ext4    defaults      0       2
/dev/sdc1       /media/movies1080             ext4    defaults      0       2
/dev/sdd2       /media/Shows2       ext4    defaults      0       2
/dev/sde1       /media/1080movies2              ext4    defaults      0       2
/dev/sdf1       /media/Shows           ext4    defaults      0       2
/dev/sdg1       /media/Torrents   ext4    defaults      0       2
/dev/sdh1       /media/Series720and1080         ext4    defaults      0       2
/dev/sdi1       /media/movies1720         ext4    defaults      0       2

I changed commit to =120 for now, once I reboot will report back on the results, but I want to do one thing at a time here. Just two days ago I uninstalled grub-pc,grub-pc-bin and grub-gfxpayload-lists and installed grub-efi-amd64 with its dependencies, thinking that since I am on a AMD processor with a UEFI bios, it would work but it would not boot (gave me a XDG_RUNTIME_DIR not set in the environment error plus a display error I forget what it said exactly, threw me into cli with UFW turned off and a bunch of other stuff turned off, asking me to run nvidia-settings --help. So (homer DOH), had to boot into livecd, copy over some files, luckily I posted my grub.cfg a few days back and I mnted then chrooted into the broken system, copied over the grub.cfg I prepared on a .txt file and reinstalled grub-pc with its dependencies via directions I found here : http://howtoubuntu.org/how-to-repair...ubuntu-live-cd. Still, when I try to install new copy of ubuntu 14.10 or vivid, my bios claims to run in insecure mode, but the box that should say try or install, is just that, a box, with nothing inside it :/ strange.

**SteveRiley** · Nov 26, 2014, 04:38 PM

Originally posted by bonkers View Post

I checked the /run/user/1000 directory and I found some interesting things in there such as: /run/user/1000/upstart/sessions/ and inside this dir is a file named 1884.session which reads

Code:

UPSTART_SESSION=unix:abstract=/com/ubuntu/upstart-session/1000/1884

, why is there an 1884, is this normal? Also a .txt file called /run/user/1000/upstart-file-bridge.1884.pid with this inside it...

I am also not allowed to change the permissions for the root directory /run/user/1000 I can untick others, ownings group, but can't apply, this is pretty much the case for anything in root. This is what it looks like:

/run is a temporary file system created in RAM by the operating system each time it boots. Various programs and services use it to store state information, caches, lock files, sockets, and data related to running processes. There is generally no good reason to tamper with the information there -- especially changing permissions. The results will be unpredictable and likely not good.

Dibl's suggestion allows you to store the Firefox cache in /run rather than on disk, because RAM is much faster than disk. This improves the performance of Firefox. You don't need to change the permissions of /run/user/1000 because your user account already owns that subdirectory. Just create a directory inside there, then change Firefox's configuration to use that directory for its cache. Because /run is transient, it will get deleted when you reboot or power down. But Firefox's configuration stays permanent; Firefox will recreate the subdirectory in /run/user/1000 next time you start it.

Originally posted by bonkers View Post

Does this look correct to you?

Yep.

**bonkers** · Nov 26, 2014, 05:06 PM

Originally posted by SteveRiley View Post

/run is a temporary file system created in RAM by the operating system each time it boots. Various programs and services use it to store state information, caches, lock files, sockets, and data related to running processes. There is generally no good reason to tamper with the information there -- especially changing permissions. The results will be unpredictable and likely not good.

Dibl's suggestion allows you to store the Firefox cache in /run rather than on disk, because RAM is much faster than disk. This improves the performance of Firefox. You don't need to change the permissions of /run/user/1000 because your user account already owns that subdirectory. Just create a directory inside there, then change Firefox's configuration to use that directory for its cache. Because /run is transient, it will get deleted when you reboot or power down. But Firefox's configuration stays permanent; Firefox will recreate the subdirectory in /run/user/1000 next time you start it.

Yep.

I found this but it says to put it into /tmp, i would like to do the way mentioned by Dibl, what would I put in fstab instead of this I found

Code:

[B]1.[/B]Edit /etc/fstab,open terminal from Applications->Accessories menu and type:[INDENT]sudo gedit /etc/fstab
[/INDENT]Add following into this file and close it.[INDENT]tmpfs /tmp tmpfs noexec,defaults,noatime 0 0
tmpfs /var/tmp tmpfs noexec,defaults,noatime 0 0
[/INDENT][B]2.[/B]Edit /etc/sysctl.conf:[INDENT]sudo gedit /etc/sysctl.conf
[/INDENT]add this line and save it:[INDENT]vm.swappiness=1
[/INDENT][B]3.[/B]
Type about:config in firefox address bar and click I'll be careful,I promise!.Right click on blank area and create a new string value called [B]browser.cache.disk.parent_directory[/B],set its value to [B]/tmp[/B]

Also in my picture, in the Advanced Permission box, who does *Others* belong too?

**bonkers** · Nov 26, 2014, 05:15 PM

Or is there a better step by step if you are not too busy, I would be very thankful of doing this. ?

**SteveRiley** · Nov 26, 2014, 06:16 PM

Originally posted by bonkers View Post

I found this but it says to put it into /tmp, i would like to do the way mentioned by Dibl, what would I put in fstab instead of this I found

You don't need anything in fstab. /run is dyamically created and mounted by the system when it boots. You can see this in the ouput of cat /proc/mounts in your post #11 here -- look at line 6. You also don't need to make any changes to /etc/sysctl.conf.

First, just create a subdirectory in the correct place under /run. I'll call this subdirectory ffcache here:

Code:

mkdir /run/user/1000/ffcache

Then, create the new string value and set its value to the directory you just created. Following the example, that will be /run/user/1000/ffcache.

Originally posted by bonkers View Post

Also in my picture, in the Advanced Permission box, who does *Others* belong too?

http://www.linux.com/learn/tutorials...le-permissions

Announcement

SSD stuck in UMDA5 mode, Timing cached reads equal to normal hard drives

SSD stuck in UMDA5 mode, Timing cached reads equal to normal hard drives

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment