[CentOS] Disk Performance Issue

mark m.roth at 5-cent.us
Mon Jun 24 15:53:35 UTC 2019


Chris Olson via CentOS wrote:
> We have a very old Dell desk top machine that has been running
> CentOS 6 for the past five years.  It received a new, 1 TB disk
> and additional memory before the OS installation.  It has been the primary
> Linux machine in our smallest and most remote field
> office.  It has been updated at least once a week and has all current dates
> installed.
>
> Boot-up this morning lasted about six times as long as usual.
> Disk access, as indicated by the disk activity light, is almost
> continuous and for extended periods of time when ever something is done
> that requires the disk.  Everything observed happens whether or not the
> machine is connected to our network. All of our files appear to be
> accessible if one is patient.
>
> One theory put forward is that some application is running that
> uses up CPU and disk bandwidth.  Another theory is that thereare disk
> errors, mostly corrected by EDCS features. We do not see any rogue
> applications and error logs show no disk issues.
>
> This is a mysterious issue that we hope to circumvent by putting
> a new disk and installing CentOS 7 from DVD.  Our hope is that the current
> disk can be mounted externally on the new CentOS system using a USB to
> SATA adapter and that data can be moved
> off of the old disk.
>
> Advice regarding this issue and any possible diagnostic methods
> will be greatly appreciated.
>
> df -k output:
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> Filesystem        1K-blocks      Used Available Use% Mounted on
> /dev/mapper/vg_delle520-lv_root
>                    51475068  12110896  36742732  25% /
> tmpfs               1928152       176   1927976   1% /dev/shm
> /dev/sda1            487652    211073    250979  46% /boot
> /dev/mapper/vg_delle520-lv_home
>                   905124888 246856176 612284356  29% /home
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>
If there's nothing in /var/log/messages, then my next trick would be to do
a smartctl -t short, and wait for the results.

Also, have you done an ipmitool sel list?

       mark



More information about the CentOS mailing list