On 29/10/2013 4:46 AM, Stan Hoeppner wrote:
On 10/28/2013 1:41 PM, Qing Chang wrote:
We have a cluster of 3 IBM servers that build the base of VMware
vSphere 5 virtual environment.
3 major servers are built that use XFS mainly because the size of
the filesystem: 64TB, 32TB and 5TB etc.
These servers run RHEL 6.4, the XFS filesystems are configured on
IBM Storwise V7000 subsystem through raw device mapping.
Two of them serve files through NFS and Samba, another one is
a SMTP/IMAP server.
Accessing files and emails does not seem to be slow. But simple commands
"ls" and "du" run very slow. For example, a first run of ls -l
/var/spool/mail could
take 8 seconds, note we have only 1360 items in this directory.
"du" would also take much longer to run than ZFS or ext3.
I could not find any direct reference of XFS possibly having problem with
these kind commands.
Could this be due to my combination of hardware and software configuration,
or is there some tuning to be done?
Many thanks,
Qing
Hello Qing,
As a Red Hat Enterprise Linux customer your contract entitles you to Red
Hat customer support. Support provided by this list is primarily for
XFS as shipped in non-proprietary 'upstream' kernels.
RHEL and SLES customers should always contact their Linux vendor for
assistance with such issues. After all that's why you paid for Linux
instead of using a free distribution.
Hi Stan,
thank you very much for your suggestion. But I do not have a support contract with RedHat,
my subscription is per University of Toronto's contract with RehHat that provides access to
the software and updates but no technical support. UofT is considering cancelling the
contract next year, so I am in a process of migrating about 20 RHEL servers to CentOS,
which is going very well.
In fact my xfs installation is NOT included in my RedHat subscription, I am using the
package that is in CentOS's repositories, I think that is what you called non-proprietary
upstream version. Following is yum list output for xfs:
xfsprogs.x86_64 3.1.1-7.el6 installed
I just would like to have an opinion from XFS gurus on this, if it is to be expected or there is
anything I can do to tune the filesystem, or maybe there is something that can be done in
the code to improve it in this respect.
I am more than happy to provide more information if anyone is interested.
Many thanks,
Qing
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
