Skip to content

Linux觀念

磁碟分割與檔案系統

磁碟分割

接近物理層的硬體上做第一層的抽象化 - 鳥哥私房菜 - 第二章、主機規劃與磁碟分割 - 在Linux系統中，每個裝置都被當成一個檔案來對待 - Linux核心支援的硬體裝置與檔名 - 整顆磁碟的第一個磁區特別的重要，因為他記錄了整顆磁碟的重要資訊 - MBR (Master Boot Record) - 第一個磁區 512bytes 會有這兩個資料 - 主要開機記錄區(Master Boot Record, MBR)：可以安裝開機管理程式的地方，有446 bytes - 分割表(partition table)：記錄整顆硬碟分割的狀態，有64 bytes - 每組分割表僅有 16bytes, 最多4組 - 主要(Primary)或延伸(Extended)分割槽 - 延伸分割最多只能1組 - GPT (GUID partition table) - 在磁區的定義上面， 大多會使用所謂的邏輯區塊位址(Logical Block Address, LBA)來處理 - 與 MBR 僅使用第一個 512bytes 區塊來紀錄不同， GPT 使用了 34 個 LBA 區塊來紀錄分割資訊 - 如何結合目錄樹的架構(邏輯上)與磁碟內(物理層面)的資料 - 掛載: 利用一個目錄當成進入點，將磁碟分割槽的資料放置在該目錄下； 也就是說，進入該目錄就可以讀取該分割槽 - 進入點的目錄我們稱為『掛載點』

檔案系統

了解磁碟分割後，引入作業系統概念，做第二層的抽象化(檔案系統) 可以將檔案系統想成作業系統要管理實際檔案時所需要的資料結構 分割槽和檔案系統的為多對多關係

``` bash= df -hT 

- [鳥哥私房菜 - 第七章、Linux 磁碟與檔案系統管理](https://linux.vbird.org/linux_basic/centos7/0230filesystem.php)
    - 磁碟分割完畢後還需要進行格式化(format)，之後作業系統才能夠使用這個檔案系統。 為什麼需要進行『格式化』呢？這是因為每種作業系統所設定的檔案屬性/權限並不相同， 為了存放這些檔案所需的資料，因此就需要將分割槽進行格式化
    - 常聽到的LVM與軟體磁碟陣列(software raid)， 這些技術可以將一個分割槽格式化為多個檔案系統(例如LVM)，也能夠將多個分割槽合成一個檔案系統(LVM, RAID)
    - 目前我們在格式化時已經不再說成針對 partition 來格式化了， 通常我們可以稱呼一個可被掛載的資料為一個檔案系統而不是一個分割槽
        - 將檔案系統與目錄樹結合的動作我們稱為『掛載』
    - 較新的作業系統的檔案資料除了檔案實際內容外， 通常含有非常多的屬性，例如 Linux 作業系統的檔案權限(rwx)與檔案屬性(擁有者、群組、時間參數等)。 檔案系統通常會將這兩部份的資料分別存放在不同的區塊，權限與屬性放置到 inode 中，至於實際資料則放置到 data block 區塊中。 另外，還有一個超級區塊 (superblock) 會記錄整個檔案系統的整體資訊，包括 inode 與 block 的總量、使用量、剩餘量等。
        - superblock：記錄此 filesystem 的整體資訊，包括inode/block的總量、使用量、剩餘量， 以及檔案系統的格式與相關資訊等；
        - inode：記錄檔案的屬性，一個檔案佔用一個inode，同時記錄此檔案的資料所在的 block 號碼；
        - block：實際記錄檔案的內容，若檔案太大時，會佔用多個 block 。
- [鳥哥私房菜 - 第 15 堂課：Linux 系統的準備 (Optional)](https://linux.vbird.org/linux_basic_train/centos8/unit15.php)

The `tmpfs` and `devtmpfs` devices are file systems in system memory. 
All files that are written to the tmpfs or devtmpfs file system disappear after a system reboot.

human-readable options
- The -h option reports in KiB (210), MiB (220), or GiB (230)
- the -H option reports in SI units: KB (103), MB (106), or GB (109)

#### /tmp
- [linux - When does /tmp get cleared? - Server Fault](https://serverfault.com/questions/377348/when-does-tmp-get-cleared)
- [How to use systemd-tmpfiles-clean to automatically remove the old logs in the directory - Red Hat Customer Portal](https://access.redhat.com/solutions/6964016)
    - [Linux下关于/tmp目录的清理规则 - 爱吃甜食的boy - 博客园](https://www.cnblogs.com/samtech/p/9490166.html)
    - /usr/lib/systemd/system/systemd-tmpfiles-clean.timer
    - /usr/lib/systemd/system/systemd-tmpfiles-clean.service
    - /usr/lib/tmpfiles.d/tmp.conf
        - [tmpfiles.d](https://www.freedesktop.org/software/systemd/man/latest/tmpfiles.d.html)
        - [Ubuntu Manpage: tmpfiles.d - Configuration for creation, deletion and cleaning of volatile and temporary](https://manpages.ubuntu.com/manpages/bionic/man5/tmpfiles.d.5.html)
- [/tmp临时目录定期清理机制 — Cloud Atlas 0.1 文档](https://cloud-atlas.readthedocs.io/zh_CN/latest/linux/redhat_linux/systemd/tmp_directory_cleanup_periodically.html)


#### mount
The /mnt directory exists for use as a temporary mount point.

If a directory to use as a mount point is not empty, then the existing files are hidden and not accessible when a file system is mounted there. 
The original files are accessible again after the mounted file system is unmounted.

It is recommended to use an unchanging device identifier to mount file systems consistently.

One stable identifier that is associated with a file system is its universally unique identifier (UUID). 
This UUID is stored in the file system superblock and remains the same until the file system is re-created.

``` bash=
# lists the full path of the device, the UUIDs and mount points, and the partition's file-system type
lsblk -fp

System shutdown and reboot procedures unmount all file systems automatically.

Unmounting is not possible when the mounted file system is in use. - For the umount command to succeed, all processes must stop accessing data under the mount point.

``` bash= mount UUID="efd314d0-b56e-45db-bbb3-3f32ae98f652" /mnt/data cd /mnt/data umount /mnt/data

lists all open files and the processes that are accessing the file system.

lsof /mnt/data cd umount /mnt/data 

### LVM
- [4.3.7. Activating and Deactivating Volume Groups Red Hat Enterprise Linux 4 | Red Hat Customer Portal](https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/4/html/cluster_logical_volume_manager/vg_activate)
- [Chapter 5. LVM Configuration Examples Red Hat Enterprise Linux 7 | Red Hat Customer Portal](https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/logical_volume_manager_administration/lvm_examples#doc-wrapper)
- [“Logical volume vg/lv contains a filesystem in use” – while removing LVM filesystem – The Geek Diary](https://www.thegeekdiary.com/logical-volume-vg-lv-contains-a-filesystem-in-use-while-removing-lvm-filesystem/)
- [Can not remove Logical Volume, message "Logical volume contains a filesystem in use" returned - Red Hat Customer Portal](https://access.redhat.com/solutions/2961861)

``` bash=
# deactive
vgchange -a n my_volume_group

# activce
vgchange -a y my_volume_group

RAID and LVM

RAID 可以將檔案系統容量增加，也有效能增加與容錯的機制，但是就是沒有辦法在既有的檔案系統架構下，直接將容量放大的機制。

LVM 最重要的任務就是進行裝置的容量放大與縮小，前提是在該裝置下的檔案系統能夠支援放大與縮小才行。 目前在 CentOS 8 上面主要的兩款檔案系統中， ext4 可以放大與縮小，但是 xfs 檔案系統則僅能放大而已。因此使用上需要特別注意。

RAID 主要的目的在效能與容錯 (容量只是附加的)，而 LVM 重點在彈性管理檔案系統 (最好不要考量 LVM 內建的容錯機制)。 若需要兩者的優點，則可以在 RAID 上面建置 LVM。

Resize

cat /proc/mdstat
mdadm --detail /dev/md126
lsof /dev/md126
Shrink xfs system on LVM on RAID

/tmp ``` bash=

enter into Resure mode after reboot

umount /tmp

remove mount point /tmp on the /etc/fstab

sed -i '/\/tmp/d' /etc/fstab

deactive and remove logic volume

lvchange -an -v /dev/rhel/tmp lvremove -vf /dev/rhel/tmp reboot

normal mode

lvcreate -L 600G -n tmp rhel mkfs.xfs -f /dev/rhel/tmp sed -i -e '$a/dev/mapper/rhel-tmp /tmp xfs defaults 0 0' /etc/fstab reboot 

### Ramdisk
- [Create a RAM disk in Linux | JamesCoyle.net Limited](https://www.jamescoyle.net/how-to/943-create-a-ram-disk-in-linux)
- [The Difference Between a tmpfs and ramfs RAM Disk | JamesCoyle.net Limited](https://www.jamescoyle.net/knowledge/951-the-difference-between-a-tmpfs-and-ramfs-ram-disk)
- [專題五 - 建立 rootfs 的無磁碟 Linux 系統環境](https://dic.vbird.tw/network_project/unit05.php)
- [ramfs, tmpfs, rootfs, initramfs的区别](https://flyflypeng.tech/%E5%86%85%E6%A0%B8/2017/06/14/ramfs-rootfs-initramfs%E7%9A%84%E5%8C%BA%E5%88%AB.html)
    - [linux - What sets the size of tmpfs? What happens when its full? - Unix & Linux Stack Exchange](https://unix.stackexchange.com/questions/205174/what-sets-the-size-of-tmpfs-what-happens-when-its-full)
        - The default is half of your physical RAM without swap

ramfs is the older file system type and is largely replaced in most scenarios by tmpfs
- it creates an in memory file system which uses the same mechanism and storage space as Linux file system cache
- it cannot be limited in size like a disk base file system which is limited by it’s capacity.
    - it will continue using memory storage until the system runs out of RAM and likely crashes or becomes unresponsive.
- you cannot see the size of the file system in df and it can only be estimated by looking at the cached entry in free command

tmpfs is a more recent RAM file system which overcomes many of the drawbacks with ramfs.
- You can specify a size limit in tmpfs which will give a ‘disk full’ error when the limit is reached.
    - This behaviour is exactly the same as a partition of a physical disk.
- tmpfs may use SWAP space.

## 網路
- [networking - what is the difference between eth1 and eno1? - Super User](https://superuser.com/questions/1053003/what-is-the-difference-between-eth1-and-eno1)

## NFS
- [nfs(5) - Linux man page](https://linux.die.net/man/5/nfs)
    - [Additional mounting considerations - Amazon Elastic File System](https://docs.aws.amazon.com/efs/latest/ug/mounting-fs-mount-cmd-general.html)
- [NFS vs CIFS - 檔案系統之間的區別 - AWS](https://aws.amazon.com/tw/compare/the-difference-between-nfs-and-cifs/)
- [Chapter 3. Mounting NFS shares Red Hat Enterprise Linux 8 | Red Hat Customer Portal](https://access.redhat.com/documentation/zh-tw/red_hat_enterprise_linux/8/html/managing_file_systems/mounting-nfs-shares_managing-file-systems#common-nfs-mount-options_mounting-nfs-shares)
- [An NFS client hangs on various operations, including "df". Hard vs Soft NFS mounts. | Support | SUSE](https://www.suse.com/support/kb/doc/?id=000020830)
- [Where are NFS defaults specified? - Unix & Linux Stack Exchange](https://unix.stackexchange.com/questions/310178/where-are-nfs-defaults-specified)
    - [fstab(5) - Linux manual page](https://man7.org/linux/man-pages/man5/fstab.5.html)
- [Chapter 110. Using Ansible to automount NFS shares for IdM users Red Hat Enterprise Linux 8 | Red Hat Customer Portal](https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/8/html/configuring_and_managing_identity_management/using-ansible-to-automount-nfs-shares-for-idm-users_configuring-and-managing-idm)
- [NFS 效能 - IBM 說明文件](https://www.ibm.com/docs/zh-tw/aix/7.3?topic=management-nfs-performance)
    - [The nfsstat -s command - IBM Documentation](https://www.ibm.com/docs/en/aix/7.3?topic=server-nfsstat-s-command)
    - [Causes of slow access times for NFS - IBM Documentation](https://www.ibm.com/docs/en/aix/7.2?topic=troubleshooting-causes-slow-access-times-nfs)
    - [Using nfsstat and nfsiostat to troubleshoot NFS performance issues on Linux | Enable Sysadmin](https://www.redhat.com/sysadmin/using-nfsstat-nfsiostat)
    - [RHEL NFS Client: slow NFS performance with \`noac\` NFS mount option - Red Hat Customer Portal](https://access.redhat.com/solutions/83723)
- [1893329 – Mounting NFS home dirs at boot fails](https://bugzilla.redhat.com/show_bug.cgi?id=1893329)
    - [Chapter 28. Mounting file systems on demand Red Hat Enterprise Linux 8 | Red Hat Customer Portal](https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/8/html/managing_file_systems/assembly_mounting-file-systems-on-demand_managing-file-systems#proc_using-systemd-automount-to-mount-a-file-system-on-demand-with-etc-fstab_assembly_mounting-file-systems-on-demand)
        - [systemd.mount](https://www.freedesktop.org/software/systemd/man/systemd.mount.html)
    - [networking - systemd: failed to mount NFS share: mount.nfs: Network is unreachable until later in boot process - Unix & Linux Stack Exchange](https://unix.stackexchange.com/questions/459731/systemd-failed-to-mount-nfs-share-mount-nfs-network-is-unreachable-until-late)
        - soft,retrans=5,timeo=30,retry=2
    - [systemd - (Ubuntu 19.04) NFS share won't automount at boot: 'mount.nfs: Network is unreachable' - Unix & Linux Stack Exchange](https://unix.stackexchange.com/questions/542925/ubuntu-19-04-nfs-share-wont-automount-at-boot-mount-nfs-network-is-unreach)
    - [fstab与systemd.mount自动挂载的一点研究和见解 · zhangguanzhang's Blog](https://zhangguanzhang.github.io/2019/01/30/fstab/#/findmnt)
    - [Reboot hangs when NFS share is mounted and lost connection to NFS server during I/O operation in Red Hat Enterprise Linux. - Red Hat Customer Portal](https://access.redhat.com/solutions/3339001)
    - [NFS hard and soft mounts - IBM Documentation](https://www.ibm.com/docs/en/spectrum-protect/8.1.9?topic=considerations-nfs-hard-soft-mounts)
- [How to log NFS related messages in "/var/log/messages"? - Red Hat Customer Portal](https://access.redhat.com/solutions/137313)
- firewall
    - [Configuring the NFS server to run behind a firewall](https://access.redhat.com/documentation/zh-tw/red_hat_enterprise_linux/8/html/deploying_different_types_of_servers/exporting-nfs-shares_deploying-different-types-of-servers#assembly_configuring-the-nfs-server-to-run-behind-a-firewall_exporting-nfs-shares)
- server is gone
    - [How to unmount NFS when server is gone? - Ask Ubuntu](https://askubuntu.com/questions/292043/how-to-unmount-nfs-when-server-is-gone)
- server not responding
    - [RHEL mount hangs: nfs: server [...] not responding, still trying - Red Hat Customer Portal](https://access.redhat.com/solutions/28211)
        - [How to increase the number of threads created by the NFS daemon in RHEL 4, 5, 6, 7 and 8? - Red Hat Customer Portal](https://access.redhat.com/solutions/2216)
    - [Correct nfs server not responding error in EFS | AWS re:Post](https://repost.aws/knowledge-center/efs-fix-nfs-server-not-responding)
        - noresvport
- NFS client hangs
    - [NFS client hang problem analysis - IBM Documentation](https://www.ibm.com/docs/en/zos/2.4.0?topic=aids-nfs-client-hang-problem-analysis)
    - [An NFS client hangs on various operations, including "df". Hard vs Soft NFS mounts. | Support | SUSE](https://www.suse.com/support/kb/doc/?id=000020830)
    - [Dell Unity: ls command hung on RHEL 8.4 client when mounting Unity NFS export via NFSv4 (User Correctable) | Dell US](https://www.dell.com/support/kbdoc/en-us/000202521/dell-emc-unity-ls-command-hung-on-rhel-8-4-client-in-a-user-correctable)
    - [Why does the command 'df' hang on a RHEL NFS client when the NFS server is unavailable? - Red Hat Customer Portal](https://access.redhat.com/solutions/23338)
    - [linux - Mount nfs hangs client - Unix & Linux Stack Exchange](https://unix.stackexchange.com/questions/404066/mount-nfs-hangs-client)


``` bash=
# on the server
nfsstat -s

# on the client
nfsstat -m
nfsstat -c
nfsiostat

``` bash=

on the server

firewall

sudo firewall-cmd --zone=public --list-all sudo firewall-cmd --add-service=nfs --permanent sudo firewall-cmd --reload sudo systemctl restart nfs-server 

Test
``` bash=
# on the client
sudo systemctl stop opt_shared.mount
sudo sed -i 's/167.22.10.1/167.22.10.2/g' /etc/fstab
sudo systemctl daemon-reload
systemctl status opt_shared.mount

TTY

Privilege Escalation

CPU

cat /proc/cpuinfo
cat /proc/cpuinfo | egrep '^processor|^physical|^core id'
numactl --hardware
lscpu
cpupower frequency-info
cpupower -c all frequency-info
lshw  -c cpu
lshw -json -c cpu

tuning

cpupower -c all frequency-set --governor performance
cpupower frequency-set -g performance

grep . /sys/devices/system/cpu/cpu*/cpufreq/scaling_driver 
grep . /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor 
grep . /sys/devices/system/cpu/cpu0/cpufreq/*
turbostat

watch -n 2 "cpupower monitor -m Mperf"
watch -n 1 cpupower frequency-info

driver - acpi-cpufreq - intel_pstate

Hardware P-States (HWP)

cpuid | grep 'hardware P-State control'

Driver

Init System

Boot method + Install target

Bootable ISO with mount ISO method

grub

Live CD for redhat => Failure - eight-livecd/ks/el8-livemedia-minimal.ks at master · alphacc/eight-livecd - lorax/docs/fedora-livemedia.ks at master · weldr/lorax - 1722595 – When building an image with livemedia-creator and a kickstart, redhat-logos must be manually added to kickstart - Kickstart for Rocky 8.7 with e16 and icewm DE’s - Rocky Linux Help & Support - Rocky Linux Forum - Directory not found after successful ISO build · Issue #1146 · weldr/lorax - Building Discovery Image on EL8 - Development - TheForeman 

# livemedia-creator
sudo dnf install lorax

# kvm
grep -E '(vmx|svm)' /proc/cpuinfo
sudo dnf install qemu-kvm libvirt virt-install
lsmod | grep kvm
sudo systemctl start libvirtd
sudo systemctl enable libvirtd

rhel9.ks 

lang en_US.UTF-8
keyboard us
timezone UTC --isUtc
auth --useshadow --passalgo=sha512
selinux --enforcing
firewall --enabled --service=ssh
services --enabled=sshd,NetworkManager
shutdown
# System bootloader configuration
bootloader --location=none
# Clear blank disks or all existing partitions
clearpart --all --initlabel
rootpw rootme
# Disk partitioning information
reqpart
part / --size=6656

%packages
@core
kernel
dracut-live
redhat-logos
memtest86+
efibootmgr
grub2-efi
grub2
shim
syslinux
%end

%post --log=/var/log/kickstart-post.log
# Optional post-install scripts
echo "LiveCD without GNOME created successfully!" > /etc/motd
%end


livemedia-creator  --iso /tmp/rhel-9.3-x86_64-dvd.iso \
  --ks=rhel9.ks  --make-iso   --releasever=9   \
  --volid=MyLiveCD  \
  --resultdir=/var/tmp/livecd  --nomacboot

livemedia-creator  --iso /tmp/rhel-9.3-x86_64-dvd.iso \
  --ks=rhel9.ks  --make-iso   --releasever=9   \
  --iso-only --iso-name=MyLiveCD.iso   --volid=MyLiveCD  \
  --resultdir=/var/tmp/livecd  --nomacboot

PXE boot

ipxe

NFS boot