Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath

Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath

[Jamie Bainbridge]

Hi all,

I was wondering if I could get a response on this bug?

 Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath
 https://bugzilla.redhat.com/show_bug.cgi?id=999761

In Bug 799842 NetApp requested that we change the default path_checker for NETAPP devices in RHEL 6.3 from directio to tur.

* Would we consider changing the default hwtable in RHEL5, considering it's in Production 2?

* Given NetApp reference the async tur improvements from Bug 760852 as part of their decision to use tur, is tur suitable for current RHEL5, or should we stick with directio?

* If tur is suitable for current RHEL5, should we change the default RHEL5 hwtable to use tur?

In Bug 799842 Comment 10 NetApp also requested we change the default features for NETAPP devices in RHEL 6.3 from "1 queue_if_no_path" to "3 queue_if_no_path pg_init_retries 50".

* pg_init_retries has been in RHEL5 since 5.2, is pg_init_retries suitable for current RHEL5 also? (this seems to be an easy "yes")

* If pg_init_retries is suitable for current RHEL5, should we change the default RHEL5 hwtable to use pg_init_retries?

Regards,
Jamie Bainbridge
Red Hat Asia Pacific

Re: Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath

[Martin George]

On 9/3/2013 5:22 AM, Jamie Bainbridge wrote:
> Hi all,
>
> I was wondering if I could get a response on this bug?
>
>   Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath
>   https://bugzilla.redhat.com/show_bug.cgi?id=999761
>
> In Bug 799842 NetApp requested that we change the default path_checker for NETAPP devices in RHEL 6.3 from directio to tur.
>
> * Would we consider changing the default hwtable in RHEL5, considering it's in Production 2?
>
> * Given NetApp reference the async tur improvements from Bug 760852 as part of their decision to use tur, is tur suitable for current RHEL5, or should we stick with directio?
>
> * If tur is suitable for current RHEL5, should we change the default RHEL5 hwtable to use tur?
>
> In Bug 799842 Comment 10 NetApp also requested we change the default features for NETAPP devices in RHEL 6.3 from "1 queue_if_no_path" to "3 queue_if_no_path pg_init_retries 50".
>
> * pg_init_retries has been in RHEL5 since 5.2, is pg_init_retries suitable for current RHEL5 also? (this seems to be an easy "yes")
>
> * If pg_init_retries is suitable for current RHEL5, should we change the default RHEL5 hwtable to use pg_init_retries?
>

On newer RHEL5 kernels like RHEL 5.9, the above changes (i.e. 'tur' path 
checker & enabling pg_init_retries) have already made it to the hwtable 
for NetApp LUNs - addressed in device-mapper-multipath-0.4.7-51.el5.

-Martin

Re: Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath

[Jamie Bainbridge]

Thanks Martin.

So it just seems the documentation in multipath.conf.defaults is outdated.

I've closed my previous bug, and logged a new bug with a patch for the docs:

 Bug 1004179 - Documented NetApp device in multipath.conf.defaults is outdated
 https://bugzilla.redhat.com/show_bug.cgi?id=1004179

Regards,
Jamie Bainbridge
Red Hat Asia Pacific

----- Original Message -----
> From: "Martin George" <marting@netapp.com>
> To: "Jamie Bainbridge" <jbainbri@redhat.com>
> Cc: "device-mapper development" <dm-devel@redhat.com>
> Sent: Tuesday, 3 September, 2013 6:52:15 PM
> Subject: Re: [dm-devel] Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath
> 
> On 9/3/2013 5:22 AM, Jamie Bainbridge wrote:
> > Hi all,
> >
> > I was wondering if I could get a response on this bug?
> >
> >   Bug 999761 - Review NetApp path_checker and features for RHEL5 mpath
> >   https://bugzilla.redhat.com/show_bug.cgi?id=999761
> >
> > In Bug 799842 NetApp requested that we change the default path_checker for
> > NETAPP devices in RHEL 6.3 from directio to tur.
> >
> > * Would we consider changing the default hwtable in RHEL5, considering it's
> > in Production 2?
> >
> > * Given NetApp reference the async tur improvements from Bug 760852 as part
> > of their decision to use tur, is tur suitable for current RHEL5, or should
> > we stick with directio?
> >
> > * If tur is suitable for current RHEL5, should we change the default RHEL5
> > hwtable to use tur?
> >
> > In Bug 799842 Comment 10 NetApp also requested we change the default
> > features for NETAPP devices in RHEL 6.3 from "1 queue_if_no_path" to "3
> > queue_if_no_path pg_init_retries 50".
> >
> > * pg_init_retries has been in RHEL5 since 5.2, is pg_init_retries suitable
> > for current RHEL5 also? (this seems to be an easy "yes")
> >
> > * If pg_init_retries is suitable for current RHEL5, should we change the
> > default RHEL5 hwtable to use pg_init_retries?
> >
> 
> On newer RHEL5 kernels like RHEL 5.9, the above changes (i.e. 'tur' path
> checker & enabling pg_init_retries) have already made it to the hwtable
> for NetApp LUNs - addressed in device-mapper-multipath-0.4.7-51.el5.
> 
> -Martin
>