YAST warning of low disk space in unrelated mounted partition with more than 800 GB free

English Applications
#1

This started happening after a disk reorg+addition I did a month ago.

Yast software manager or updater donwnload the rpms/deltas, OK, but give this warning before installing each package if I don’t tick “don’t show this warning again”:

“The disk space in partition /home/shares/Gravados/Series\040Concluidas/ is almost full. Continue installation?”

This partition is sdf1. It, plus sde1 and scd1, are mounted below sdd1. All partitions occupy a full 2TB disk each, ext4-formatted.

The relevant lines of fstab are shown below, with UUIDs substituted for sdXX for better understanding. the \040 code translate to a blank space in the mount directory. Yast partitioner would choke on them (or the accents), but the names could be directly edited in fstab and work fine after rebooting:


...
sdd1    /home/shares                 ext4       noatime,data=writeback,acl,nodiratime 1 2
...
sde1    /home/shares/Gravados/Series ext4       noatime,data=writeback,acl,user_xattr,nodiratime 1 2
sdc1    /home/shares/Gravados/Series\040em\040Inglês       ext4       noatime,data=writeback,acl,user_xattr,nodiratime 1 2
sdf1    /home/shares/Gravados/Series\040Concluidas/ ext4       noatime,data=writeback,acl,user_xattr,nodiratime 1 2

sda and sdb are system and data drives, not linked to the others.

Any idea how to identify the cause of problem?

Thanks.

P.S.: I know that using accented letters and spaces in mount names is not good practice, but I’ve been using it for years without issue.

#2

If something is complaining about a file system being (almost) full, I would check that with

df -h
#3

There’s a lot of free space there:

> df -h
Sist. Arq.      Tam. Usado Disp. Uso% Montado em
/dev/sda6        20G   11G  8,0G  58% /
devtmpfs        7,9G  4,0K  7,9G   1% /dev
tmpfs           7,9G   96K  7,9G   1% /dev/shm
tmpfs           7,9G  2,3M  7,9G   1% /run
tmpfs           7,9G     0  7,9G   0% /sys/fs/cgroup
/dev/sdb1       100G   30G   71G  30% /windowsD
/dev/sdb6       990M  162M  761M  18% /var/log
/dev/sdb5       976M  346M  563M  39% /tmp
/dev/sda2        90G   23G   68G  25% /windowsC
/dev/sda7       118G   69G   44G  62% /home
/dev/sdb3       718G  544G  139G  80% /home/arquivos
/dev/sdb2        99G   54G   40G  58% /home/trabalho
/dev/sdd1       1,8T  790G  951G  46% /home/shares
/dev/sdc1       917G  666G  251G  73% /home/shares/Gravados/Series em Inglês
/dev/sdf1       1,8T 1022G  **812G**  56% /home/shares/Gravados/Series Concluidas
/dev/sde1       1,8T  888G  946G  49% /home/shares/Gravados/Series

Just tested transferring about 30GB of files to it, no issues. Could it be something related to the /tmp mount?

#4

Could be the /tmp partition seems a bit small to me lots of things use /tmp

#5

I agree. Bit small. I wonder why you have a separate file system for /tmp. That does not make using free space for short times easier. You must have a reason for it, because it is not in the default offering during installation.

#6

I didn’t want to put /tmp in the SSD, and from searching around, 1GB seemed like a good size, after changing K3B tmp directory to somewhere else.

BUT, I cleaned up tmp, rebooted (perhaps not necessary) and did an update (today’s krb5 security update) without getting the warning message, so it seems that was it.

On a related note, before systemd there was a setting in Yast’s etc/config editor like CLEAR_TMP_ON_BOOT, but it’s not there anymore. How can I do this? I’m not familiar with cron, and it might not work here because of systemd too…

#7
  1. Make a copy of /usr/lib/tmpfiles.d/tmp.conf to /etc/tmpfiles.d/tmp.conf. The copy will overrule the original at runtime, while the original can still be updated by any software install/update.
  2. Edit the copy using
man tmpfiles.d

as documentation. I use this

# Clear tmp directories separately, to make them easier to override
# Eigen policy: maak /tmp en /var/tmp leeg.
D /tmp 1777 root root 1d
D /var/tmp 1777 root root 1d

# Exclude namespace mountpoints created with PrivateTmp=yes
x /tmp/systemd-private-*
x /var/tmp/systemd-private-*
X /tmp/systemd-private-*/tmp
X /var/tmp/systemd-private-*/tmp

, but that is on 13.1. On 13.2 there is a better solution that does it on boot.

#8

I found it on a 13.2 system.

When you do change

D /tmp 1777 root root 1d
D /var/tmp 1777 root root 1d

into

D! /tmp 1777 root root 1d
D! /var/tmp 1777 root root 1d

the action is only done at boot. Which might be more safe (and also might be what you want when your system is booted e.g. each day).

It is all in

man tmpfiles.d
#9

Thanks for the help, hcvv, I appreciate it.

According to man there is a fifth argument (called Argument, what a surprise!), and empty/default arguments seem to require a dash. The man page doesn’t say so, I’m presuming because of the example in it. OTOH maybe trailing arguments don’t need to be declared. Anyway I’ll try this, and see how it goes:

# Clear tmp directory at boot
D! /tmp 1777 root root 10d -
#10

Duh, scratch that. The settings in /usr/lib/tempfiles.d/tmp.conf don’t use it for the fifth argument, so it’s obviously not necessary. What was I thinking?

#11

And the warning message is back :frowning:

/tmp - 904 MB free
/var/log - 756 MB free
/var/tmp - 7.9 GB free
…/Series\040Concluidas - 811 GB free

Suggestions, anyone?

#12

Note that for a user they get that wanring at about 95% full I think The reason is to allow some space to let root in to fix things.

What are you doing when you get the warning?? You know that /tmp is temp so whatever is triggering the warning may have come and gone by the time you get around to looking at the space.

Are you still using a 1 gig /tmp??

BTW we perfer to see actual output if possible in code tags (# in editor’s bar)

#13

On 2015-03-22 19:06, brunomcl wrote:
>
> And the warning message is back :frowning:
>
> /tmp - 904 MB free

That’s too low.

Anyway…
Try umounting “/home/shares/Gravados/…” before starting YaST…

–
Cheers / Saludos,

Carlos E. R.
(from 13.1 x86_64 “Bottle” at Telcontar)

#14

Good idea. Meanwhile, it’s not for lack of inodes either:

:~> df -i
Sist. Arq.        Inodes IUsado     ILivr IUso% Montado em
/dev/sda6        1310720 334330    976390   26% /
devtmpfs         2048244    762   2047482    1% /dev
tmpfs            2050314      8   2050306    1% /dev/shm
tmpfs            2050314   1154   2049160    1% /run
tmpfs            2050314     15   2050299    1% /sys/fs/cgroup
/dev/sdb1       74032004   7960  74024044    1% /windowsD
/dev/sdb5          65536     78     65458    1% /tmp
/dev/sdb6          65408    149     65259    1% /var/log
/dev/sda2       71274960 120559  71154401    1% /windowsC
/dev/sda7        7864320  89495   7774825    2% /home
/dev/sdb3       47816704  86523  47730181    1% /home/arquivos
/dev/sdb2        6553600  91939   6461661    2% /home/trabalho
/dev/sdd1      122101760  22406 122079354    1% /home/shares
/dev/sdc1       61054976   3275  61051701    1% /home/shares/Gravados/Series em Inglês
/dev/sdf1      122101760   6719 122095041    1% /home/shares/Gravados/Series Concluidas
/dev/sde1      122101760   6326 122095434    1% /home/shares/Gravados/Series
#15

There’s, plenty of space everywhere, see https://forums.opensuse.org/showthread.php/506145-YAST-warning-of-low-disk-space-in-unrelated-mounted-partition-with-more-than-800-GB-free?p=2700646#post2700646

/tmp now has 906 MB free. I doubt that’s the problem:

:~> df -h
Sist. Arq.      Tam. Usado Disp. Uso% Montado em
/dev/sda6        20G   11G  8,0G  58% /
devtmpfs        7,9G     0  7,9G   0% /dev
tmpfs           7,9G   92K  7,9G   1% /dev/shm
tmpfs           7,9G  2,3M  7,9G   1% /run
tmpfs           7,9G     0  7,9G   0% /sys/fs/cgroup
/dev/sdb1       100G   30G   71G  30% /windowsD
/dev/sdb5       976M  3,5M  906M   1% /tmp
/dev/sdb6       990M  166M  758M  18% /var/log
/dev/sda2        90G   23G   68G  25% /windowsC
/dev/sda7       118G   56G   57G  50% /home
/dev/sdb3       718G  558G  124G  82% /home/arquivos
/dev/sdb2        99G   54G   40G  58% /home/trabalho
/dev/sdd1       1,8T  804G  937G  47% /home/shares
/dev/sdc1       917G  667G  250G  73% /home/shares/Gravados/Series em Inglês
/dev/sdf1       1,8T 1022G  812G  56% /home/shares/Gravados/Series Concluidas
/dev/sde1       1,8T  891G  943G  49% /home/shares/Gravados/Series

Yes, shouldn’t be a problem from what I’ve researched around. And the warning is about another partition, with 812 GB free.

Yeah, that’s just me being lazy, sorry. :wink:

#16

Hello,

Same problem here (and here), with an external usb drive whose label contained a blank.

The cause seems to be the blank in the mount point. Remove it and it should be ok.

There’s something wrong in YaST or libzypp or elsewhere. \040 is escaped somewhere (\040), then the mount point is lost and YaST gets a size of 0 available bytes used (on a total of 0) when trying to get disk usage. Not related with a real lack of space.

It really sounds like a bug to me. On the other hand, blanks in mount points are not recommended. Anyway I haven’t found the time to make a good bug report yet… feel free to do it if you can reproduce this “bug” :slight_smile:

Antoine

#17

Hi Antoine,

Thank you for the insight.

I’ve also noticed that the warning only happens during installation through Yast’s software manager. It does not appear with Yast Online Update nor with packagekit/Software Updater frontend, that works very well with KDE4 I must say.

I’m not sure a bug report is worth the time, as the developers will probably say I shouldn’t be using spaces in the volume name anyway, and the bug is innocuous. I’ll try to give it a try, if you know what I mean :slight_smile:

Regards,

Bruno

#18

Reported, FWIW: https://bugzilla.opensuse.org/show_bug.cgi?id=925088