Benefits of rebuilding the kernel?

Question

MySQL

The opinion of the collective mind is required.
There is a new server. It has centos(x86_64). The server is planned under the mysql database. Is it worth rebuilding the kernel? Will it give a performance boost?
Thank you.

Answer 1

On it centos

Is it worth rebuilding the kernel?

NO!
Install gent if you want to "rebuild".

Answer 2

I can’t confirm with my own experience, but I think that there won’t be a significant increase in performance, it’s much more useful to tweak my.cnf to fit your needs.
But I'll be happy to see other answers in this thread.

Answer 3

By the forces of redhat, everything was tested for a specific patchset for a specific version of the kernel with specific options, and all the documentation was written for it. By digging around, there is a small chance to grab underground knocks that no one but you has. Works - do not touch.

Answer 4

It only makes sense to rebuild the kernel to include missing features/drivers. For example, to include additional required targets in the packet filter. The rest is a harmful and dangerous occupation.

Answer 5

What exactly in the kernel do you want to change? Without a clear understanding of what the options being changed are responsible for, you can do anything. Reassembly for the sake of reassembly is definitely not needed. And it certainly won’t help for mysql from scratch, it’s more useful to do tuning by other means, as they wrote above.

Answer 6

And why are you not satisfied with the default kernel? More precisely: what SPECIFICALLY does it not suit you? After all, when you rebuild it, do you plan to tweak something and turn it on / off? Or is it so simple, to collect ...?

Answer 7

You can tune the timer, change the preemptive model, patch at worst, roll, but the performance will hardly change if you don't patch it.

Answer 8

The author mentioned BSD, and so, if you follow binary updates in it, then you can’t rebuild the kernel in principle. Also in any Linux, if updates, then only the default kernel. In a word, on important servers, it is impossible to rebuild the kernel. But if the server is used for development or tests, then you can try to twist something

Answer 9

My experience shows that if the question is put like this, then it is not necessary, but most likely it will only get worse :) compilation parameters will be changed, it does not bring profit, but it creates problems on an equal footing (the simplest of them is the complexity with automatic updating).

Answer 10

No need to rebuild anything on the centos, especially the kernel. In the same place 2.6.18 with a bunch of backports. Do you have anything to add?
For a bulkhead, this is in gentoo.
In addition, you will not help the muscle with this operation. Absolutely. The only way to help the muscle in the kernel is to change the io scheduler. But it can be done without rebuilding.

Answer 11

I would achieve speed by tuning my.cnf and sysctl.
This will give much more benefit than building the kernel.