Subject: Re: create/symlink failed, no inodes free
To: Ross Harvey <ross@ghs.com>
From: Mocha <netbsd_alpha@yahoo.com>
List: port-alpha
Date: 05/09/2000 18:57:04
At 04:39 PM 5/9/00 -0700, Ross Harvey wrote:
> > From: Mocha <netbsd_alpha@yahoo.com>
> >
> > system: dec alpha pws 500au / 1gig RAM / 20 gig SCSI HD
> > os: netbsd 1.4.2 / default install kernel
> > action: untar-ing source sets
> >
> > 1. tar -xzvf gnusrc.tgz -C / (this went fine)
> > 2. tar -xzvf pkgsrc.tgz -C / (got error below)
> >
> > here is a strange error i got:
> > ---cut---
> > /: create/symlink failed, no inodes free
> > tar: Could not create directory pkgsrc/www/navigator3 : No such file or
> > director
> > y
> > pkgsrc/www/navigator3/CVS/
> >
> > /: create/symlink failed, no inodes free
> > tar: Could not create directory pkgsrc/www/navigator3/CVS : No such
> file or
> > dire
> > ctory
> > pkgsrc/www/navigator3/CVS/Root
> >
> > /: create/symlink failed, no inodes free
> > ---cut---
> >
> > how come the system's default settings don't have enough inodes to
> handle a
> > simple operation like that? or is something else i should be looking into?
>
>Looks like you tried to fit source trees into the root FS .. they usually
>go into /usr/... or on some FS that has enough room. It was just a tossup
>whether you ran out of inodes first or space first.
>
>Maintaining a small root FS is usually a good practice, although it is in
>less favor than it used to be. The theory is, it's easier to backup as an
>image for emergency recovery, it's accessed less and so is less likely to
>be inconsistent in a crash, filling up root or /usr won't stop ops on /usr
>or root, and if something does go wrong it's vastly easier to restore /usr
>with a working root, or to regenerate root from the distribution and drop
>in the local config automatically saved for you in /var/backups. There are
>probably other reasons too...
alpha500# df -k
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/sd0a 127431 62708 58351 51% /
/dev/sd0d 1601733 426740 1094906 28% /usr
looks like i still had space on / and it was getting close to being done
(well at the W section). how does the system determine if there is not
enough inodes given that there is enough space left?
i usually set root for about 100 mb. each new disk i add, i make partion a
mirror the original root. so if the original go bad, i can still boot from
the other disks and rebuilt faster.
i was actually untar-ing source tree into /usr. i just assumed that all the
tar balls had the highest parent tree as usr/.../... they all did except
pkgsrc.tgz. so i just let it continue and was planning to move it after
knowing that i had enough space on /.
on a side note, each time i add more HDs, i make a new swap partition equal
to that of my max ram. then add that new swap partition to the old swap
pool. am i wasting disk space? i just have this fear of one day running out
of swap space. so for this system, each new HD has 1gig set aside for swap
(max ram is 1gig right now).