I will be taking this host down for maintenance tomorrow. I don't know if
EDEX keeps any processed HDF5 files.
What I wonder is if 55,000,000 is a sufficient amount of inodes for basic
EDEX installations.
I have had problems before on a filesystem which we moved to zfs. After
that, we no longer saw any I/O hangs. I'm hoping a good purge, tomorrow,
on the /data_store/ directory will get the host back to "acting normal"
(aka no more high loads and I/O hangs). The Geosciences department which
asked for access to an EDEX server isn't yet using it in the classroom!
On Oct 22, 2015 3:12 PM, "Justin Balbierer" <jbalbier@xxxxxxxxx> wrote:
> Hi Donna -
>
> Have you checked the configuration of your purge files?
>
> If the issue with the raw data or processed hdf5 files?
>
> Justin
>
> On Tue, Oct 20, 2015 at 12:57 PM, Donna Cote <d-cote@xxxxxxxx> wrote:
>
>> Our EDEX server is reaching
>> <
>> 20% of its inodes limit (
>> limit
>> is
>> set to over
>> 55,000,000). I am trying to find out if this is expected or
>> if it is
>> unusual.
>>
>> Maybe, simply reducing the number of files (hoping to remove everything
>> over 10 days old) will "solve" this problem.
>>
>>
>> This filesystem has a size of 836Gb, and is holding all of the
>> data_store LDM data that the LDM server saves.
>>
>> [root@awips2 ~]# df -h /awips2
>> Filesystem Size Used Avail Use% Mounted on
>> /dev/mapper/vg_awips2-lv_awips2
>> 836G 794G 16M 100% /awips2
>>
>> [root@awips2 ~]# df -i /awips2
>> Filesystem Inodes IUsed IFree IUse% Mounted on
>> /dev/mapper/vg_awips2-lv_awips2
>> 55672832 45367382 10305450 82% /awips2
>> [root@awips2 ~]#
>>
>> Please email me if you have any help for me with this.
>>
>> Donna Cote
>> Texas A&M University
>>
>> _______________________________________________
>> awips2-users mailing list
>> awips2-users@xxxxxxxxxxxxxxxx
>> For list information, to unsubscribe, or change your membership options,
>> visit: http://www.unidata.ucar.edu/mailing_lists/
>>
>>
>
