SCIENTIFIC-LINUX-USERS Archives

February 2007

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Michael Mansour <[log in to unmask]>
Reply To:
Michael Mansour <[log in to unmask]>
Date:
Sat, 10 Feb 2007 18:50:58 +1000
Content-Type:
text/plain
Parts/Attachments:
text/plain (58 lines)
Hi,

I've built two new SL4.4 servers with /var an LV of 2Gb size and ext3.

After running through some server testing, I found I quickly ran out of inodes
on /var with 64% of the filesystem used but 100% of inodes used.

df -i shows:

                      262144  256844    5300   98% /var

I've been using Linux now for 14 years and this is the first time I've ran out
of inodes.

These are currently test servers so it's no big deal, but if these servers
were in production and this happens, what can be done?

The mke2fs man page shows:

       -i bytes-per-inode
              Specify  the  bytes/inode ratio.  mke2fs creates an inode for
every bytes-per-inode bytes of space
              on the disk.  The larger the bytes-per-inode ratio, the fewer
inodes will be created.  This  value
              generally  shouldnât  be  smaller than the blocksize of the
filesystem, since then too many inodes
              will be made.  Be warned that is not possible to expand the
number of inodes on a filesystem after
              it is created, so be careful deciding the correct value for this
parameter.

and:

       -N number-of-inodes
              overrides the default calculation of the number of inodes that
should be reserved for the filesys-
              tem (which is based on the number of blocks and the
bytes-per-inode ratio).  This allows the  user
              to specify the number of desired inodes directly.

So it seems that the 262144 value is inadequate and I'll have to specify more
when building the filesystem, but I would think it should be an easy task to
add inodes when in a production environment using ext2online or similar, which
doesn't seem the case.

It seems if in production, the only option would be to backup the filesystem,
reformat it with a higher -N value, restore the data. Not a good solution at
all as production servers almost never get rebooted my end (they're HA clustered).

Does anyone have any other options? or has anyone experienced this problem before?

I'm really after knowledge to know if inodes can be given to a filesystem
while the filesystem is online and mounted.

Thanks.

Michael.

ATOM RSS1 RSS2