Core dump and implications on the size of the partitions /var and swap

[aleunix]

4.8 - How should I partition my disk?

Swap: In addition to swap, this partition is also used for storing core dumps after system crashes, so it is suggested that the swap space (if set up at all) be bigger than the largest amount of RAM you are likely to ever install on the machine.

/var/crash: If the system panics, it will attempt to save a core dump in the swap partition before rebooting. This core dump will then be saved to /var/crash upon reboot, so /var will need at least as much free space as the system has RAM for this to work automatically.

I don't want save an eventual core dump of the system, so i would like known if using partitions more smaller can cause some problems or if the system, simply, don't save nothing given that understand that there isn't sufficient space.

[jggimi]

When a core dump is taken:

1. The kernel saves the core dump and control information to swap space.
2. On boot, rc(8) executes savecore(8) to copy any core dumps to /var/swap.

If you want diagnostic information captured when the kernel panics, you will need sufficient swap space so that the kernel can capture the dump. The savecore(8) program will not copy the dump from swap to /var/crash if there is insufficient space.

Since you don't want to capture diagnostic information on any failures, you need not retain space in /var for a core dump.

[aleunix]

If even the swap partition is smaller, what happen during a core dump?

[jggimi]

The default sysctl settings will cause the kernel to enter ddb(4) on a panic or other form of kernel crash. No core dump is taken from ddb unless explictly requested. See crash(8) for additional information.

[aleunix]

crash - system failure and diagnosis (web man of the command).
Thanks.