System thinks its running out of space
Warren Post
along the lines of:
The volume "Filesystem root" has only 194.8 MB disk space remaining
But df disagrees:
$ df
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 7.7G 3.4G 4.0G 46% /
/dev/sda8 139G 93G 46G 67% /home
/dev/sda5 2.1G 68M 1.9G 4% /opt
none 499M 88K 499M 1% /tmp
/dev/sda6 2.1G 113M 1.9G 6% /var/www
Searching the web for the text of the pop-up finds nothing useful. System
mail has nothing relevant in it.
Any thoughts on how I can diagnose this problem?
Jim Beard
mount |grep root
My wag is you have a partition with name or mount-point /root
that has only 194.8 MB disk space remaining.
if the pop-up said The volume "Filesystem /" has only 194.8 MB
disk space, this would indicate something different, of course.
And if / and root are synonyms as far as GNOME is concerned, I am
far off track, of course.
Cheers!
jim b.
--
UNIX is not user unfriendly; it merely
expects users to be computer-friendly.
Warren Post
> mount |grep root
>
> My wag is you have a partition with name or mount-point /root that has
> only 194.8 MB disk space remaining.
A good thought, but alas, no mount point named root:
[warren@verdi ~]$ mount | grep root
[warren@verdi ~]$ mount
/dev/sda1 on / type ext4 (rw)
none on /proc type proc (rw)
/dev/sda8 on /home type ext4 (rw)
/dev/sda5 on /opt type ext4 (rw)
none on /tmp type tmpfs (rw)
/dev/sda6 on /var/www type ext4 (rw)
none on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)
gvfs-fuse-daemon on /home/warren/.gvfs type fuse.gvfs-fuse-daemon
(rw,nosuid,nodev,user=warren)
Jim Beard
What is on /dev/sda7 ?
And maybe /dev/sda9 ?
If you have blkid installed,
blkid /dev/sda7
Warren Post
> What is on /dev/sda7 ?
> And maybe /dev/sda9 ?
$ blkid /dev/sda7
/dev/sda7: UUID="dd9da04a-21c4-4d8b-8bff-9d153648c08d" TYPE="swap"
The same command for sda9 comes back empty, so I guess that means there
is no sda9.
If I knew what program was calling the pop-up, I'd run it manually from a
terminal and see if anything interesting goes to stdout.
Bit Twister
> On Mon, 10 Jan 2011 08:59:54 -0500, Jim Beard wrote:
>
>> What is on /dev/sda7 ?
>> And maybe /dev/sda9 ?
>
> $ blkid /dev/sda7
> /dev/sda7: UUID="dd9da04a-21c4-4d8b-8bff-9d153648c08d" TYPE="swap"
>
> The same command for sda9 comes back empty, so I guess that means there
> is no sda9.
Yep. Once in awhile I find this command helpful if you have a
partition number greater than 9.
blkid | sort -V
Aragorn
identifying as Warren Post wrote...
Now that others have thrown in their WAGs, here's mine...
It may be possible for "df" to report more available free space on the
filesystem than there is, if the number of available free inodes on the
system is only enough to cover for the 194.4 MB of diskspace that you
see being reported as still available.
One could of course beg the interesting question how come you've
wasted/used up so many inodes on the root filesystem then. Well,
considering that you have both "/usr" and "/var" as part of the root
filesystem, you can start by looking there, and most notably,
in "/var". Look at "/var/log" and friends.
Many small files take up many inodes, and there are only so many inodes
on an ext3/4 filesystem you can use, regardless of how many diskspace
you still leave available.
Hope this helps. ;-)
--
*Aragorn*
(registered GNU/Linux user #223157)
Peter D.
in the Usenet newsgroup alt.os.linux.mandriva
Aragorn wrote:
If that is the problem then giving df the "-i" option will display it.
--
Peter D.
Sig goes here...
Warren Post
> It may be possible for "df" to report more available free space on the
> filesystem than there is, if the number of available free inodes on the
> system is only enough to cover for the 194.4 MB of diskspace that you
> see being reported as still available.
>
> One could of course beg the interesting question how come you've
> wasted/used up so many inodes on the root filesystem then. Well,
> considering that you have both "/usr" and "/var" as part of the root
> filesystem, you can start by looking there, and most notably, in "/var".
> Look at "/var/log" and friends.
Right clicking on /var in Thunar and selecting Properties tells me it
contains 4371 items, totalling 182.7 MB used. /usr is more interesting
with 140081 items, totalling 4.6 GB used -- and df reports less space
used for all of the root partition:
$ df /dev/sda1
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 7.7G 3.5G 3.9G 48% /
Tediously doing the same for every directory in the root partition I get:
/bin 114 items, totalling 13.8 MB
/boot 34 items, totalling 25.4 MB
/dev 802 items, totalling 2.2 GB
/etc 4464 items, totalling 188.3 MB
/initrd 1 item, totalling 108 B
/lib 3945 items, totalling 79.3 MB
/lost+found 0 items, totalling 0 B
/mnt 1 item, totalling 4.0 kB
/root 88 items, totalling 261.4 kB
/sbin 328 items, totalling 55.0 MB
/sys 13335 items, totalling 551.4 MB
/tmp 57 items, totalling 134.8 kB
/usr 140081 items, totalling 4.6 GB
/var 4371 items, totalling 182.7 MB
(An aside: right clicking in Thunar does the job, but how do I get this
information from the command line?)
All this totals to 167621 items and 7.89 GB, which would explain the
warning messages and suggest I am about to be in big trouble.
I spent a little time poking around /usr, but since I don't know what is
supposed to be in there, I can't tell what is wrong. I do know that my
previous Linux installs have all had root partitions using 4 GB or less
total, so something appears to be amiss.
Anticipating the worst, I've confirmed that my nightly backups are good.
But what can I do to bring /usr under control?
Warren Post
> If that is the problem then giving df the "-i" option will display it.
Here's the output of df -i, with plain vanilla df included for comparison:
[warren@verdi ~]$ df
Filesystem Size Used Avail Use% Mounted on
/dev/sda1 7.7G 3.5G 3.9G 48% /
/dev/sda8 139G 98G 41G 71% /home
/dev/sda5 2.1G 68M 1.9G 4% /opt
none 499M 124K 499M 1% /tmp
/dev/sda6 2.1G 113M 1.9G 6% /var/www
/dev/sr0 413M 413M 0 100% /media/
Drakbackup_20110111_040201
[warren@verdi ~]$ df -i
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/sda1 501K 149K 352K 30% /
/dev/sda8 8.8M 36K 8.8M 1% /home
/dev/sda5 131K 15 131K 1% /opt
none 125K 72 125K 1% /tmp
/dev/sda6 131K 3.5K 128K 3% /var/www
/dev/sr0 0 0 0 - /media/
Drakbackup_20110111_040201
Jim Beard
<snip>
> $ df /dev/sda1
> Filesystem Size Used Avail Use% Mounted on
> /dev/sda1 7.7G 3.5G 3.9G 48% /
>
> Tediously doing the same for every directory in the root partition I get:
>
> /bin 114 items, totalling 13.8 MB
> /boot 34 items, totalling 25.4 MB
> /dev 802 items, totalling 2.2 GB
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
How do you manage to get 2.2 GB into /dev?
And only 802 items?
> /etc 4464 items, totalling 188.3 MB
> /initrd 1 item, totalling 108 B
> /lib 3945 items, totalling 79.3 MB
> /lost+found 0 items, totalling 0 B
> /mnt 1 item, totalling 4.0 kB
> /root 88 items, totalling 261.4 kB
> /sbin 328 items, totalling 55.0 MB
> /sys 13335 items, totalling 551.4 MB
> /tmp 57 items, totalling 134.8 kB
> /usr 140081 items, totalling 4.6 GB
> /var 4371 items, totalling 182.7 MB
>
> (An aside: right clicking in Thunar does the job, but how do I get this
> information from the command line?)
>
> All this totals to 167621 items and 7.89 GB, which would explain the
> warning messages and suggest I am about to be in big trouble.
Note that /sys is nothing more than pointers to data contained in
various places. It occupies no disk space to speak of. Initrd
likewise is empty except during boot.
du and du -s are useful to checking size of directories.
ls -laR /directory |wc -l
will provide a count of directories+files+links.
How to get&keep /usr under control is another issue entirely.
My system is definitely heavily loaded with just about everything
except server software, and often there is a 32-bit and a 64-bit
version of libraries and some packages with both installed.
Add-in desktop backgrounds do take up a good bit of space, as
does multilingual support for Chinese, Japanese, and Vietnamese.
[root@jdb usr]# du -s /usr
7.3G /usr
To put this in context,
[root@jdb usr]# du -s /bin /boot /dev /etc /home /lib /lib64 /opt
/root /sbin /usr /var
6.9M /bin
49M /boot
23M /dev
117M /etc
4.1G /home
233M /lib
17M /lib64
7.2M /opt
122M /root
15M /sbin
7.3G /usr
325M /var
[root@jdb usr]# for i in /bin /boot /dev /etc /home /lib /lib64
/opt /root /sbin /usr /var
> do echo -e $i \\c ; ls -laR $i |wc -l
> done
/bin 121
/boot 71
/dev 2330
/etc 7344
/home 22208
/lib 30334
/lib64 404
/opt 463
/root 8422
/sbin 260
/usr 471859
/var 5912
My solution was to buy a WD Caviar Terabit HD, set it up as 100
GB partitions, and use a partition for each install, with my
music and genealogy stuff in separate partitions accessed from
whichever partition happens to be booted at the time.
Warren Post
> How do you manage to get 2.2 GB into /dev? And only 802 items?
Thunar says that. Nautilis (again doing the right click thing) says 499
items, totaling 1.2 GB. But the command line gives me more believable
results:
[root@verdi warren]# du -s /dev
15M /dev
[root@verdi warren]# ls -laR /dev/ | wc -l
1157
I guess the lesson here is don't trust GUI tools blindly. So to repeat my
earlier check, but using du:
# du -s /bin/ /boot/ /dev/ /etc/ /lib/ /root/ /sbin/ /tmp/ /usr/ /var/
7.0M /bin/
9.7M /boot/
15M /dev/
135M /etc/
74M /lib/
396K /root/
21M /sbin/
84K /tmp/
3.0G /usr/
221M /var/
...totaling 3.57 GB, which is consistent with what df has been reporting
all along.
So I'm tempted to conclude that the pop-up messages that prompted me to
start this thread are spurious and I should suppress them. Anyone think
otherwise?
David W. Hodgins
> So I'm tempted to conclude that the pop-up messages that prompted me to
> start this thread are spurious and I should suppress them. Anyone think
> otherwise?
Looking that way. What's the output of "cat /proc/mounts" first
though?
Regards, Dave Hodgins
--
Change nomail.afraid.org to ody.ca to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)
Warren Post
> On Tue, 11 Jan 2011 23:06:38 -0500, Warren Post
> <inv...@invalid.invalid> wrote:
>
>> So I'm tempted to conclude that the pop-up messages that prompted me to
>> start this thread are spurious and I should suppress them. Anyone think
>> otherwise?
>
> Looking that way. What's the output of "cat /proc/mounts" first though?
$ cat /proc/mounts
rootfs / rootfs rw 0 0
/dev/root / ext4 rw,relatime,barrier=1,data=ordered 0 0
/dev /dev devtmpfs rw,relatime,size=507408k,nr_inodes=126852,mode=755 0 0
/proc /proc proc rw,relatime 0 0
/sys /sys sysfs rw,relatime 0 0
none /dev/pts devpts rw,relatime,mode=620 0 0
none /dev/shm tmpfs rw,relatime 0 0
/dev/sda8 /home ext4 rw,relatime,barrier=1,data=ordered 0 0
/dev/sda5 /opt ext4 rw,relatime,barrier=1,data=ordered 0 0
none /tmp tmpfs rw,relatime 0 0
/dev/sda6 /var/www ext4 rw,relatime,barrier=1,data=ordered 0 0
none /proc/sys/fs/binfmt_misc binfmt_misc rw,relatime 0 0
/proc/bus/usb /proc/bus/usb usbfs rw,relatime,devgid=43,devmode=664 0 0
gvfs-fuse-daemon /home/warren/.gvfs fuse.gvfs-fuse-daemon
rw,nosuid,nodev,relatime,user_id=500,group_id=500 0 0
David W. Hodgins
> $ cat /proc/mounts
> /dev/root / ext4 rw,relatime,barrier=1,data=ordered 0 0
I don't see anything strange. Given the df output, I'd either
ignore it, or report it as a bug at
https://qa.mandriva.com/index.cgi
I have not seen a similar problem here, but I rarely use Gnome.
Aragorn
identifying as Warren Post wrote...
> On Tue, 11 Jan 2011 22:04:16 +0100, Aragorn wrote:
>
>> It may be possible for "df" to report more available free space on
>> the filesystem than there is, if the number of available free inodes
>> on the system is only enough to cover for the 194.4 MB of diskspace
>> that you see being reported as still available.
>>
>> One could of course beg the interesting question how come you've
>> wasted/used up so many inodes on the root filesystem then. Well,
>> considering that you have both "/usr" and "/var" as part of the root
>> filesystem, you can start by looking there, and most notably, in
>> "/var". Look at "/var/log" and friends.
>
> Right clicking on /var in Thunar and selecting Properties tells me it
> contains 4371 items, totalling 182.7 MB used. /usr is more interesting
> with 140081 items, totalling 4.6 GB used -- and df reports less space
> used for all of the root partition:
>
> $ df /dev/sda1
> Filesystem Size Used Avail Use% Mounted on
> /dev/sda1 7.7G 3.5G 3.9G 48% /
>
> Tediously doing the same for every directory in the root partition I
> get:
>
> /bin 114 items, totalling 13.8 MB
> /boot 34 items, totalling 25.4 MB
> /dev 802 items, totalling 2.2 GB
Whoa! 2.2 GB for "/dev"? That can't be right... 8-|
> /etc 4464 items, totalling 188.3 MB
> /initrd 1 item, totalling 108 B
> /lib 3945 items, totalling 79.3 MB
> /lost+found 0 items, totalling 0 B
> /mnt 1 item, totalling 4.0 kB
> /root 88 items, totalling 261.4 kB
> /sbin 328 items, totalling 55.0 MB
> /sys 13335 items, totalling 551.4 MB
> /tmp 57 items, totalling 134.8 kB
> /usr 140081 items, totalling 4.6 GB
> /var 4371 items, totalling 182.7 MB
>
> (An aside: right clicking in Thunar does the job, but how do I get
> this information from the command line?)
man du
;-)
> All this totals to 167621 items and 7.89 GB, which would explain the
> warning messages and suggest I am about to be in big trouble.
Well, the population of "/dev" normally lives on a tmpfs, and thus in
virtual memory - i.e. RAM plus swap - so I'm really curious as to what
could possibly be using up 2.2 GB in there. And furthermore, how come
you're not seeing this reflected in RAM or swap usage.
> I spent a little time poking around /usr, but since I don't know what
> is supposed to be in there, I can't tell what is wrong.
"/usr" is the hierarchy that contains the multi-user software. It's
essentially a read-only filesystem - by which I do not mean that it is
necessarily mounted read-only, but it could just as well be, because it
does not require being written to. There is a directory "/usr/tmp"
which some legacy applications may still write to, but in all GNU/Linux
distributions I've seen so far ever since my very first Mandrake 6.0
installation, "/usr/tmp" is a symbolic link to "/var/tmp".
The chances that there is anything wrong in your "/usr" hierarchy are
very small. Even if you regularly download new updates and/or other
packages and you keep the cache with the packages, then that is not
where they go - they go under "/var/urpmi" instead on Mandriva.
> I do know that my previous Linux installs have all had root partitions
> using 4 GB or less total, so something appears to be amiss.
My root partition is a *lot* smaller than that... :p
Filesystem Size Used Avail Use% Mounted on
/dev/hda2 393M 179M 214M 46% /
/dev/hda1 197M 38M 159M 20% /boot
/dev/hda3 9.8G 2.5G 7.3G 26% /usr
/dev/hda6 746M 33M 714M 5% /opt
/dev/hda7 2.0G 355M 1.6G 18% /var
/dev/hda8 298M 33M 266M 11% /usr/local
/dev/hda9 79G 4.3G 75G 6% /home
/dev/hda10 21G 3.2G 18G 16% /srv
none 1014M 52K 1014M 1% /tmp
> Anticipating the worst, I've confirmed that my nightly backups are
> good. But what can I do to bring /usr under control?
"/usr" is not your problem. "/dev" is. Or at least, in theory, because
since the population of "/dev" lives on a tmpfs in all systems which
use the udev system, such a large chunk of storage would be nibbled off
of your available RAM and/or swap.
Warren Post
> Whoa! 2.2 GB for "/dev"? That can't be right... 8-|
Good call; it wasn't right. That information came from right clicking on
/dev in Thunar and selecting "Properties". Doing it the right way gives a
sane result:
# du -s /dev
15M /dev
Moral: Don't trust GUI tools blindly; learn how to do everything from the
command line. I know you already know that, Aragorn, I'm just muttering
to myself here.