Le 23/10/2012 14:47, Les Mikesell a écrit :
> On Tue, Oct 23, 2012 at 3:23 AM, Frédéric Massot
> <frederic@...> wrote:
>>> Just curious: does the content you back up consist mostly of tiny
>>> files without much duplication? It seems odd to run out of inodes
>>> while still having substantial disk space.
>> I used the command "df -i" on different servers to see those who ate a
>> lot of inode, and it seems that it is the web server with Apache cache
>> Is that one inode is used by one directory?
> Each file and directory will use one, except that hardlinked files
> share their inode.
>> I'll use the "-t" option of htcacheclean to remove empty directory, and
>> see if it makes a difference.
>> On a web server most affected:
>> $ df -h
>> Filesystem Size Used Avail Use% Mounted on
>> /dev/vda1 92G 52G 36G 60% /
>> $ df -i
>> Filesystem Inodes IUsed IFree IUse% Mounted on
>> /dev/vda1 6111232 4391859 1719373 72% /
>> And the folder "/var/cache/apache2/mod_disk_cache" contains 2956954
>> directories to 13 GB. The limit of the total disk cache size (-l option)
>> is 300 MB ?!
> If it is just a cache, it might be reasonable to exclude it from
> backups. The contents would probably have expired by the time you
> could restore anyway, or already have fresh copies reloaded from the
> original source.
Yes, I put that directory in the list of exclusions.
For the health of this web server, adding the "-t" htcacheclean option
was good. After one day the number of inode consumed decreased from 72%
| FRÉDÉRIC MASSOT |
| http://www.juliana-multimedia.com |
| mailto:frederic@... |
| +33.(0)220.127.116.11.94 +33.(0)18.104.22.168.69 |