linux_wiki:kernel_tuning

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

linux_wiki:kernel_tuning [2016/03/18 23:08]
billdozor [Kernel Tuning] 		linux_wiki:kernel_tuning [2019/05/25 23:50]
Line 1: Line 1:
-====== Kernel Tuning ====== 
- 
-**General Information** 
- 
-Kernel tuning using sysctl.conf 
- 
-**Checklist** 
-  * Distro(s): Any distro, but many examples are done from Red Hat based systems 
- 
----- 
- 
-====== Hugepages ====== 
- 
-Problem: Unable to access a VM's console, and it shows out of memory errors on boot. 
- 
-Solution: Reduce the value of "vm.nr_hugepages" to less than 1/2 of total memory in MBs. 
- 
-/etc/sysctl.conf 
-<code bash> 
-vm.nr_hugepages = 1024 
-</code> 
-1 hugepage = 2 MBs, so the above is equal to 2GB 
- 
-Explanation: Hugepages that is configured to be greater than or equal to total installed memory causes memory starvation during boot due to no memory available for normal use. 
- 
-Alternative config location: Hugepages could also be configured at /boot/grub/grub.conf 
- 
----- 
- 
-==== HugePage Bug ==== 
- 
-On some kernels, khugepaged can start running at 100% CPU utilization. This is typically seen on systems that have process/memory intensive processes. 
- 
-Other symptoms include: 
-  * Normal commands hang (w, uptime, ps) 
-  * Processes showing above 100% CPU (ie sometimes up to 1,300%) 
- 
-Details 
-  * Known Affected kernels 
-    * CentOS 6: 2.6.32-431.el6.x86_64 
-  * Reference bug report: https://bugzilla.redhat.com/show_bug.cgi?id=879801 
- 
-=== Workaround Fix === 
- 
-Disable hugepage defragmenting: 
-<code bash> 
-echo never > /sys/kernel/mm/redhat_transparent_hugepage/defrag 
-</code> 
- 
-Add new cron entry 
-<code bash> 
-vim /etc/cron.d/hugepage_defrag 
- 
-# Disable kernel huge page defrag due to bugzilla bug: 879801 
-@reboot root /bin/echo never > /sys/kernel/mm/transparent_hugepage/defrag 
-</code> 
- 
-=== Permanent Fix === 
- 
-Ultimately, the permanent fix is to update the kernel to a newer version. 
- 
-TODO: Will add confirmed kernel versions that this bug is fixed in. 
- 
----- 
- 
-====== Swappiness ====== 
- 
-Swappiness controls how likely the kernel is to move processes out of memory and onto swap disk space. 
- 
-The setting is from 0 to 100. 
-  * 0 = Very aggressively avoid swapping for as long as possible 
-    * High risk of OOM killing from memory and I/O pressure 
-  * 10 = Red Hat recommended for Oracle databases 
-  * **60 = Linux default** 
-  * 100 = Aggressively swap from memory to disk 
- 
-Check the setting: 
-<code bash> 
-cat /proc/sys/vm/swappiness 
-60 
-</code> 
- 
-To change the setting: edit sysctl.conf, add a line, save, and re-read the config: 
-<code bash> 
-vim /etc/sysctl.conf 
-vm.swappiness=10 
-:wq 
-sysctl -p 
-</code> 
- 
-More details about OOM: https://owlbearconsulting.com/doku.php?id=linux_wiki:oom_killer 
- 
----- 
- 
-====== Page Cache ====== 
- 
-Tuning how often cached pages in memory are flushed to disk. 
- 
-Check current settings: 
-<code bash> 
-sysctl -a | grep dirty 
-</code> 
- 
-Background (async) pagecache flushing: 
-<code bash> 
-vm.dirty_background_ratio = 3 
-</code> 
-  * Explanation: Start flushing pages to disk asynchronously when pagecache is equal to 3% of memory. 
-  * Default value = 10 
- 
-Foreground (sync) pagecache flushing: 
-<code bash> 
-vm.dirty_ratio = 15 
-</code> 
-  * Explanation: Sync flush when pagecache is 15% of total memory. This blocks ANY process from using I/O when this happens. If this process takes longer than 120 seconds, kernel panics can happen. 
-  * Default value = 20 
- 
----- 
- 
-====== File Descriptors ====== 
- 
-Viewing and setting the system wide file descriptors. 
- 
-**View system wide file descriptor information** 
-<code bash> 
-sysctl fs.file-nr 
-fs.file-nr = 6144 0 809542 
-</code> 
- 
-The three numbers returned: 
-  * 6144 = used file descriptors 
-  * 0 = allocated but not used 
-  * 809542 = system max 
- 
- 
-**Change System FD Limits** 
- 
-Edit /etc/sysctl.conf 
-<code bash> 
-fs.file-max = 810542 
-</code> 
-Example above increases system wide max by 1,000. 
- 
-More details, including per user settings: https://owlbearconsulting.com/doku.php?id=linux_wiki:file_descriptors 
- 
----- 
- 
-====== Kernel Panic ====== 
- 
-Reboot a system after a 10 seconds of kernel panic 
-<code bash> 
-kernel.panic = 10 
-</code> 
- 
----- 
  
  • linux_wiki/kernel_tuning.txt
  • Last modified: 2019/05/25 23:50
  • (external edit)