Sol2.6 realloccg: fs full - is my FS boo-boo?

Sol2.6 realloccg: fs full - is my FS boo-boo?

Post by Andres Kroonm » Tue, 06 Jan 1998 04:00:00



 Hi,

 some time ago asked about the meaning of a message:

     NOTICE: realloccg: file system full

 while there was at least 2GB os free space on volume
 and at least 1M free files available.

 FS was made to reserve 0% of space to root, later
 changed that to 1%. No diff, non-root-user gets
 FS full long before it actually is. I expect root
 to have the same limit, but I'd not test it, as
 pretty big population of users depend on this box
 to run smoothly.

 the volume itself is 17GB in size and has full-limit
 at about 14.5 GB (ie at ca 85%)

 Now, lately made a fsck on it:
** /dev/rdsk/c0d1s2
** Last Mounted on /www/disk1
** Phase 1 - Check Blocks and Sizes
** Phase 2 - Check Pathnames
** Phase 3 - Check Connectivity
** Phase 4 - Check Reference Counts
** Phase 5 - Check Cyl groups
978428 files, 14600754 used, 2905688 free
 (2534944 frags, 46343 blocks, 429496719.6% fragmentation)

now the fragmentation figure makes me real nervous. What
the heck is going on with this volume?

"realloccg" makes me think of something like cylinder group,
can anyone spread some light on this?

 thanks,

 
 
 

Sol2.6 realloccg: fs full - is my FS boo-boo?

Post by Karsten Thygese » Wed, 07 Jan 1998 04:00:00


Andres>  Hi, some time ago asked about the meaning of a message:

Andres>      NOTICE: realloccg: file system full

Andres>  while there was at least 2GB os free space on volume and at
Andres>  least 1M free files available.

Andres>  FS was made to reserve 0% of space to root, later changed
Andres>  that to 1%. No diff, non-root-user gets FS full long before
Andres>  it actually is. I expect root to have the same limit, but I'd
Andres>  not test it, as pretty big population of users depend on this
Andres>  box to run smoothly.

I had the same problem (using Solaris-2.5) on my 8GB newsspool disk a
few weeks ago. After help from this newsgroup and investigation with
"typfs -v", it turned out, that one of the cylinder groups was full
(the others where not).

I used minfree at 3% at that time.

The solution was to use tunefs to adjust minfree to 7-10% and remove
some files. As this was a newsspool area, all files will be rewritten
within quite short time and the problem went away.

I was advised to newer use minfree less than 8%, so I use 10% now
instead.

Hope you could use this advice!

Best,
Karsten

--
M.Sc.E.E, Sys/Net-administrator, BOFH, Aalborg University, Denmark
http://www.kom.auc.dk/~karthy/ for more information.
SunSITE Denmark Manager (http://sunsite.auc.dk/)
# $Id: .signature,v 1.1 1997/06/18 08:30:00 karthy Exp $

 
 
 

1. root "no shell" boo-boo

Subject: root "no shell" boo-boo

The scenario:

A friend has a brand new Sparc Station 5.  After a few hours of
installing Solaris from the CD-ROM, the first action while logged
in as root was to edit /etc/passwd and change root's shell from
/sbin/sh to /sbin/csh.  Oops.

Now, on login as root, the error message, "no shell" appears and
returns to the login prompt.

The problem, of course, is that csh is not in the /sbin directory.
No other logins were created.  Bummer.  What can they do to
fix this, besides installing the entire Solaris CD-ROM from
scratch?


2. Debian Potato

3. Which versions?

4. boo-boo follow-up

5. Which News Software is Best?

6. Boo-boo in XFree86 1.3 linux binaries (Cirrus driver)

7. INSTALL JDK ON MKLINUX R1

8. su, pam, Red Hat 7.0: I made a boo-boo

9. Root boo boo

10. The server made a boo boo error, please help

11. Upgrade Boo Boo....

12. Kyocera FS 1600 or FS 1600+ Printer drivers ( FS-1600 FS-1600+)