Kernel changes and patch install before reboot

[Gareth Eagar]

Hi All,

Please could you offer me some advice about whether it is safe to manually make changes to the kernel (system tuning changes) and to then install patches BEFORE doing a reboot.

IE, I would like to limit the process to one reboot to cover both the kernel changes and patch install. I am presuming that the patch install will require a reboot and am concerned that somehow the changes the patch install does will overwrite or cause to be lost the kernel changes I make manually.

Can this be done safely?

Regards,


Gareth

[melvyn burnard]

well it would help to know which patches, as that would tell us if they will reboot the system
In eihter case, make sure you have a good backup, preferably a make_recovery just in case.
I also like to suggest you try a mk_kernel before any of this, so you at least know hte kernel builds ok before starting work.
Too many times I have seen kernel build failures htat were already there before someone loaded patches.

[Ravi_8]

Hi

PHSS and PHCO patches doesn't need reboot, whereas some of PHNE and all PHKL patches needs reboot.
There is no harm in reooting the machine twice.
even if you make kernel changes manually (using kmtune or by modifying /stand/system) system needs a reboot for them to be effective

[Pete Randall]

Gareth,

While I can understand your desire to minimize downtime, I would prefer to do these tasks separately. If you should find that there are problems building the kernel, or re-booting the system, you'll have a much better idea what the likely cause was.


Pete

[Ross Zubritski]

I must consur with Pete. One step at in time in making environment changes. Makes it much easier to backtrack in case something goes awry.


Regards.

RZ

[Marco Santerre]

Gareth,

If you look at the patch description it will tell you whether you need to reboot or not. That being said, I would still do two reboots. The reason behind my logic is I'd like to make sure that the kernel changes I make will not affect the kernel changes that the patch may be doing. Moreover, I'd probably do my kernel changes last just to make sure they take effect the way I want them to. That last sentence is how *I* would do it, there's no valid reason not to do it the other way around.