Be less nice #798
Be less nice #798
Conversation
…rformance. Signed-off-by: DL6ER <dl6er@dl6er.de>
Signed-off-by: DL6ER <dl6er@dl6er.de>
| @@ -367,6 +370,20 @@ void read_FTLconf(void) | |||
| if(buffer != NULL && strcasecmp(buffer, "false") == 0) | |||
| config.block_esni = false; | |||
|
|
|||
| // NICE | |||
| // Shall we change the nice of the current process? | |||
| // defaults to: -10 (can be disabled by setting value to -999) | |||
There was a problem hiding this comment.
Is there a reason for the default to be -10? This is a question we will be asked many times.
There was a problem hiding this comment.
Is there a reason for the default to be
-10?
No
Quoting man sched:
With the advent of the CFS scheduler in kernel 2.6.23, Linux adopted an algorithm that causes relative differences in nice values to have a much stronger effect. **In the current implementation, each unit of difference in the nice values of two processes results in a factor of 1.25 in the degree to which the scheduler favors the higher priority process.** This causes very low nice values (+19) to truly provide little CPU to a process whenever there is any other higher priority load on the system, and makes high nice values (-20) deliver most of the CPU to applications that require it (e.g., some audio applica‐ tions).
This means that a niceness of -10 will result in roughly 9x more cycles given to pihole-FTL compared to other processes when they are all asking for 100% CPU usage.
…change the nicencess. Signed-off-by: DL6ER <dl6er@dl6er.de>
|
Out of curiosity: Will I see DNS server performance increase in my small home network where my CPU is already dying of boredom? Or rephrased: how many clients do I need before I'll profit from this PR? |
|
@yubiuser It will not affect you at all when you run Pi-hole as the only process on your machine. However, when installing Pi-hole next other compute-intense, i.e., CPU-hungry, processes like a database or e-mail server, maybe a VPN service, etc., then this change will ensure that Pi-hole receives a greater share of the available computing power (it will be less equally distributed among the processes, hence, less "nice" to them). |
|
Oh. Merge conflicts |
Signed-off-by: DL6ER <dl6er@dl6er.de>
|
@PromoFaux Merge conflict resolved |
|
This pull request has been mentioned on Pi-hole Userspace. There might be relevant details there: https://discourse.pi-hole.net/t/pi-hole-5-1-released/35577/1 |
|
This pull request has been mentioned on Pi-hole Userspace. There might be relevant details there: https://discourse.pi-hole.net/t/raspberry-pi-4-hohe-last-keine-domain-auflosung/37152/4 |
See pi-hole/FTL#798 Without this capability, users will see `WARNING: Required Linux capability CAP_SYS_NICE not available` in FTL log
See pi-hole/FTL#798 Without this capability, users will see `WARNING: Required Linux capability CAP_SYS_NICE not available` in FTL log Signed-off-by: Craig Younkins <cyounkins@gmail.com>
By submitting this pull request, I confirm the following:
How familiar are you with the codebase?:
10
Set nice value of pihole-FTL to increase DNS server performance. The new niceness default to
-10(lower is less nice = getting more CPU cycles for our own). The value can be changed through a config option. Changing the niceness can be avoided by settingNICE=0(this will set0as niceness, which is the default) orNICE=-999(this will skip trying to set the niceness, altogether).If the user is not allowed to change their niceness, FTL will log:
If the user specified an invalid niceness, it will be clamped to the according extremum (
+/- 20):If everything is fine, the output is:
TODO (pi-hole/core): Add
CAP_SYS_NICEto FTL so it is allowed to change its own niceness when when running as userpihole. Without this, onlyrootcan change the niceness of processes.