Remount option in Service guard

[Mike_305]

Hello,

We have several Service guard pkgs and we are trying to set "nosuid" option in control scripts.

I have set up that in the service guard file but I do not want to recycle the pkg and implement the "nosuid".

LV[5]="/dev/SG-FS/lvol1"; FS[5]="/ora/lvol1"; FS_TYPE[5]="vxfs"; FS_MOUNT_OPT[5]="-o largefiles,delaylog,nosuid"; FS_UMOUNT_OPT[5]=""; FS_FSCK_OPT[5]=""

I know this can be done on non-service guard file system using following syntax but not able to do in with SF file systems.

mount -p /test >>> To check

mount -o remount /oracle_test >>> to apply option

The question that I have is can this be done without briging down the pkg?

Appreciate your help in advance.

Regards,

MJ

[Steven E. Protter]

Shalom MJ,

The answer depends here.

If the volume group is mounted in shared mode, I believe mount change like this will fail.

However this is complex enough that you probably want to try it out in the lab with your cluster configuration.

SEP

[Hasan Atasoy]

hi ,

write temporarily fs entries to /etc/fstab.


remount will work.


Hasan.

[Mike_305]

Hello,

This on my TEST server, and it's only one FS that I have issue with. ON my DEV system we lots of FS that I need this to be set and goal is t oavoide reboot of SG pkgs.

I have also try to put the entry in FS tab and it won't work.

Any idea from any guru's.

Thanks for your help in advance.

Regards,

MJ

[Mike_305]

Steven,

The VG is in exclusive mode.

Regards,

MJ

[Patrick Wallek]

>>I have also try to put the entry in FS tab and it won't work.

What won't work? What error do you get? What, exactly, did you add to /etc/fstab? What command did you then type at the prompt? What output did you receive?

[James R. Ferguson · Accepted solution]

Hi MJ:

I think that this is doable. Make your package changes so that the next time the package control scripts are run, you have the options you want.

Using 'mount -o remount ...' does *not* examine '/etc/fstab'. You *must* specify all of the mount options you want on the commandline when you issue the 'mount'.

Regards!

...JRF...

[Mike_305]

Hello,

Sorry me again.

In FSTAB this is what I put in. All I am trying to do is set "nosuid" option to this file system.

/dev/SG-FS/lvol1 /ora/lvol1 vxfs largefiles,delaylog,nosuid 0 2

After I did this;

mount -o remount /ora/lvol1 >>> No ERROR

The ERROR I got was without entry in fstab - was either ignored or not found in /etc/fstab.

then check using "mount -p"

mount -p | grep /ora/lvol1 >> "nosuid" options is not set.

mount -o remount has worked for me on non-SG file systems. The idea behinf "remount" is to be able to do this without un-mounting the FS and be able to refresh the options with new setting.

I have to do this without rebooting the PKG. Is their any way this can be done, if the answer is no then I guess I have no choice.

Thanks for all your help.

I hope I answer your questions.

Regards,

MJ

[Mike_305]

Bingo!!!!

Thanks guys, just figured it out. This is what I needed.

mount -o remount,nosuid /ora/lvol1 >>>> Works on SF File system.

AND I was doing;

mount -o remount /ora/lvol1 >>> Does not work on SG but work on NON-SG File systems.

Now to do the SG File system. I had to do the entry in FS, control file and run following command.

mount -o remount,nosuid /ora/lvol1

[James R. Ferguson]

Hi (again) MJ:

> In FSTAB this is what I put in. All I am trying to do is set "nosuid" option to this file system.

Re-read what I wrote about putting *all* the mount options that you want as the arguments to '-o remount'. The '/etc/fstab' is *not* consulted when you issue this. See the 'mount_vxfs(1M) manpages.

Regards!

...JRF...

[Mike_305]

JRF,

Your little hint was good for my peanut brain. LOL

You *must* specify all of the mount options you want on the commandline when you issue the 'mount'.

Regards,

MJ

[Mike_305]

Agreed JRF but when I did not had the entry in FSTAB it was giving me this error.

Command RAN: mount -o remount,nosuid /ora/lvol1

ERROR I got - mount: /ora/lvol1 was either ignored or not found in /etc/fstab

Not sure wht, I had to put the entry in FSTAB. Without that won't work.

Thanks for all your help. Enjoy the points.

Regards,

MJ

[James R. Ferguson]

HI (again) MJ:

> Agreed JRF but when I did not had the entry in FSTAB it was giving me this error.

Had you specified the 'special' or 'directory' argument in the commandline mount() the *absence* of an entry in 'fstab' should not have been a problem:

Given the syntax:

mount [-F vxfs] [-eQrV] [-o specific_options] { special | directory }

...the manpages note that "If you omit either special or directory, mount attempts to determine the missing value from an entry in /etc/fstab".

Regards!

...JRF...

[Mike_305]

JRF,

No I was only putting following.

mount -o renount,nosuid /ora/lvol1.

May be that's why.

Cool, it's always plsure to learn form the guru's.

Regards,

MJ

[Mike_305]

Dog on it, I am bad when I am rushing myself. I wish they make spell check button on the ITRC FORUMS.

LOL,

Appreciate everyone's help and thanks JRF.

Regards,

MJ

[Steven E. Protter]

Shalom MJ,

I think your plan will work as well.

The remount will probably work, but do expect some interesting results on volume groups activated in shared mode.

It would clarify things for me if you explained why you need to check a mount and then change options.

Would it not be better to simply mount the file system in the chosen mode?

There is a risk of the package going down or failing over if you make a mistake. I'd be in contact with my user community before making changes in this regards.

SEP

[James R. Ferguson]

HI MJ:

I meant to write:

# mount[-Fvxfs] [-eQrV] [-ospecific_options] special directory

....that is without the curly braces and the alternation of 'special' and 'directory' in the case we are discussing.

/* No points for this correction */

Regards!

...JRF...

[Mike_305]

Hello,

We are dong this on TEST systems.

Thanks for everyone's help.

Enjoy the points.

Regards,

MJ