Re: IO Freeze ON RHEL 6.6 and hence high log file sync

Hi Stefan,
Can you pls elaborate "what does DM-multipather say".

I dont think there are any failing paths because the storage logs and/or dmesg would have reported that. Neither did.

This happened in January then directly in April. So we moved the application to DR site and next day it happened at DR site too. So I am least suspecting it to be hardware issue. We switched back to the primary site (after doing the huge page changes) and since then it has been fine.

Can you pls elaborate on "blktrace to drill down". Can it help when the problem is not happening? For eg it happened for a few minutes (3 minutes) on those 3 occassions and then it has been fine.

Many thanks

On 19 April 2018 at 12:54, Stefan Koehler <contact_at_soocs.de> wrote:

> Hey Purav,
> what does the DM-multipather say? Do you have failing paths? Otherwise you
> can use blktrace to drill-down.
>
> Best Regards
> Stefan Koehler
>
> Independent Oracle performance consultant and researcher
> Website: http://www.soocs.de
> Twitter: _at_OracleSK
>
> > Purav Chovatia <puravc_at_gmail.com> hat am 18. April 2018 um 19:25
> geschrieben:
> >
> > We are on RHEL 6.6 running a 2-node Oracle 12c RAC.
> >
> > All of a sudden, for 2-3 minutes, DATADG (asm diskgroup containing all
> data files) freezes for 2-3minutes with iostat showing 100% utilisation,
> very high svctm and very high write_await. But the r/s, w/s rsec/s and
> wsec/s is 0. Meaning no IO but still disk fully busy. So it’s like a freeze.
> >
> > Has anybody experienced this and can somebody help as to what could be
> the reason?
> >
> > Transparent huge pages were not disabled and general huge pages were not
> configured. Now both done.
> >
> > Many thanks
>

--
http://www.freelists.org/webpage/oracle-l