| 1 |
-----BEGIN PGP SIGNED MESSAGE----- |
| 2 |
Hash: SHA1 |
| 3 |
|
| 4 |
Brandon Adams wrote: |
| 5 |
| I woild assume that you would configure / build a new kernel for each |
| 6 |
| hardware spec in your farm in your test environment, verify there are |
| 7 |
| no glitches and then distibute the .config file to all servers and |
| 8 |
| cron a kernel build / installation. |
| 9 |
|
| 10 |
I'd say that depends on your idea of kernel building. |
| 11 |
We prefer a general kernel with static drivers for crucial |
| 12 |
hardware/option and module drivers for non-crucial hardware/options |
| 13 |
|
| 14 |
Combined with module autoloading this allows for a flexible system with |
| 15 |
little overhead. |
| 16 |
|
| 17 |
I know that there are several people in the security community that |
| 18 |
advertise disabling module-loading, however consider the problems you're |
| 19 |
in if someone is actually able to load modules on one of your servers. |
| 20 |
|
| 21 |
| The reboot required for the servers would then be done during that |
| 22 |
| server's maintenance window. |
| 23 |
|
| 24 |
We're currently researching if we can reduce the maintenance down-time |
| 25 |
for kernel reloading by using kexec. On large memory servers and |
| 26 |
scsi/raid controllers bios re-initialization can easily take up to 10 |
| 27 |
minutes. (that's pre-bootloader) |
| 28 |
|
| 29 |
Ramon |
| 30 |
-----BEGIN PGP SIGNATURE----- |
| 31 |
Version: GnuPG v2.0.7 (GNU/Linux) |
| 32 |
|
| 33 |
iD8DBQFHsWqtwiVM6CtDHQ0RAujCAJkB4lBFyxLTfIcGI1Iwfx1k8b5AOgCbBrrk |
| 34 |
SJIlqHBVcFsfx4VVcFoEdRU= |
| 35 |
=ZdJY |
| 36 |
-----END PGP SIGNATURE----- |
| 37 |
-- |
| 38 |
gentoo-server@l.g.o mailing list |
Archives