问题：如下例是下载 Debian bookworm 的网络启动文件，再在当前 Linux 系统里用 kexec 启动。

wget http://ftp.cn.debian.org/debian/dists/bullseye/main/installer-amd64/current/images/netboot/debian-installer/amd64/initrd.gz
wget http://ftp.cn.debian.org/debian/dists/bullseye/main/installer-amd64/current/images/netboot/debian-installer/amd64/linux

kexec --command-line="priority=critical mirror/http/hostname=ftp.cn.debian.org interface=auto  language=en country=US locale=en_US.UTF-8 keymap=us modules=network-console network-console/password=taobao925 network-console/password-again=taobao925 DEBIAN_FRONTEND=text"  --initrd=initrd.gz linux

但在国内阿里云 ECS 里实践操作是失败的，sshd 没成功连接。查看 /var/log/syslog 日志，有如下的信息：

Feb  7 08:06:39 main-menu[259]: (process:1632): udhcpc: lease of 172.30.118.192 obtained from 172.30.127.253, lease time 315360000
Feb  7 08:06:39 main-menu[259]: INFO: Falling back to the package description for brltty-udeb
Feb  7 08:06:39 main-menu[259]: INFO: Falling back to the package description for brltty-udeb
Feb  7 08:06:39 main-menu[259]: INFO: Menu item 'choose-mirror' selected
Feb  7 08:06:39 anna-install: Queueing udeb apt-mirror-setup for later installation
Feb  7 08:06:39 choose-mirror[1719]: DEBUG: command: wget --no-verbose http://deb.debian.org/debian/dists/oldstable/Release -O - | grep -E '^(Suite|Codename|Architectures):'
Feb  7 08:08:50 choose-mirror[1719]: DEBUG: command: wget --no-verbose http://deb.debian.org/debian/dists/stable/Release -O - | grep -E '^(Suite|Codename|Architectures):'
Feb  7 08:08:53 choose-mirror[1719]: DEBUG: command: wget --no-verbose http://deb.debian.org/debian/dists/bookworm/Release -O - | grep -E '^(Suite|Codename|Architectures):'
Feb  7 08:08:59 choose-mirror[1719]: DEBUG: command: wget --no-verbose http://deb.debian.org/debian/dists/testing/Release -O - | grep -E '^(Suite|Codename|Architectures):'
Feb  7 08:11:09 choose-mirror[1719]: DEBUG: command: wget --no-verbose http://deb.debian.org/debian/dists/unstable/Release -O - | grep -E '^(Suite|Codename|Architectures):'
Feb  7 08:13:20 choose-mirror[1719]: INFO: suite/codename set to: stable/bookworm
Feb  7 08:13:20 choose-mirror[1719]: DEBUG: command: wget --no-verbose http://deb.debian.org/debian//dists/bookworm/main/binary-amd64/Release -O - | grep ^Architecture:
Feb  7 08:13:21 anna-install: Queueing udeb bookworm-support for later installation
Feb  7 08:13:21 main-menu[259]: (process:1712): failed: Connection timed out.
Feb  7 08:13:21 main-menu[259]: (process:1712): failed: Network is unreachable.
Feb  7 08:13:21 main-menu[259]: (process:1712): 2024-02-07 08:08:53 URL:http://deb.debian.org/debian/dists/stable/Release [149229/149229] -> "-" [1]
Feb  7 08:13:21 main-menu[259]: (process:1712): 2024-02-07 08:08:59 URL:http://deb.debian.org/debian/dists/bookworm/Release [149229/149229] -> "-" [1]
Feb  7 08:13:21 main-menu[259]: (process:1712): failed: Connection timed out.
Feb  7 08:13:21 main-menu[259]: (process:1712): failed: Network is unreachable.
Feb  7 08:13:21 main-menu[259]: (process:1712): failed: Connection timed out.
Feb  7 08:13:21 main-menu[259]: (process:1712): failed: Network is unreachable.
Feb  7 08:13:21 main-menu[259]: (process:1712): 2024-02-07 08:13:21 URL:http://deb.debian.org/debian//dists/bookworm/main/binary-amd64/Release [116/116] -> "-" [1]
Feb  7 08:13:21 main-menu[259]: INFO: Falling back to the package description for brltty-udeb
Feb  7 08:13:21 main-menu[259]: INFO: Falling back to the package description for brltty-udeb
Feb  7 08:13:21 main-menu[259]: INFO: Menu item 'download-installer' selected
Feb  7 08:13:21 net-retriever: regular: dists/bookworm/Release /tmp/net-retriever-1776-Release
Feb  7 08:15:31 anna[1774]: failed: Connection timed out.
Feb  7 08:15:31 anna[1774]: failed: Network is unreachable.
Feb  7 08:17:14 anna[1774]: WARNING **: bad d-i Packages file

大概是尝试连接默认的 deb.debian.org 仓库站失败了。搜索得， deb.debian.org 站是由 Fastly 提供的 CDN 站。估计，deb.debian.org 在国内访问有问题。但为什么启动参数里指定的 mirror/http/hostname=ftp.cn.debian.org 无效呢？查看 preseed 文件和说明举例，知道镜像站的配置是有几个参数必须同时指定，如 mirror/country=manual 和 mirror/http/directory=/debian。那么修改一下启动的参数，如 kexec --command-line="priority=critical mirror/country=manual mirror/http/hostname=ftp.cn.debian.org mirror/http/directory=/debian interface=auto locale=en_US.UTF-8 keymap=us modules=network-console network-console/password=taobao925 network-console/password-again=taobao925 DEBIAN_FRONTEND=text" --initrd=initrd.gz linux 就能达到安装时使用 ftp.cn.debian.org 的目的了。

参考：

• https://www.debian.org/releases/stable/i386/apbs02.zh-cn.html#idm3216
• https://www.debian.org/releases/stable/i386/apbs04.en.html
• https://www.debian.org/releases/bookworm/example-preseed.txt
• https://wiki.debian.org/DebianInstaller/FAQ#Q:_How_do_I_copy_d-i_logfiles_to_a_remote_host.3F
• https://deb.debian.org/

需求：将 LiveCD grml.iso 保存在 ntfs 分区上，使用 grub4dos 来引导，加载 grml.iso 文件。

本例，硬盘是 MBR 分区，只有一个 NTFS 分区，已经安装有 Windows 22H2。grml.iso 保存在 C:\iso\grml.iso。

尝试：

1. 先安装 grub4dos, 将 grub4dos ，用 bcdedit 将 grub4dos 添加到启动菜单中

2. 这里的 C:\menu.lst 文件内容，其中 findiso 参数是让它能在设备中查找iso文件，toram=grml.squashfs 是避免 ntfs-3g 自动将 ntfs 分区挂载为“可读写”，导致重启或关机时会挂死

   title grml
   find --set-root --ignore-floppies --ignore-cd /iso/grml.iso
   map /iso/grml.iso (0xff)
   map --hook
   set opts=" toram=grml.squashfs dns=8.8.8.8 ip=192.168.1.108::192.168.1.1:255.255.255.0:testbox:eth0:off ssh=mima "
   kernel (0xff)/boot/grml/vmlinuz apm=power-off boot=live live-media-path=/live/grml/ bootid=grml001 findiso=/iso/grml.iso  nomce net.ifnames=0 %opts%
   initrd (0xff)/boot/grml/initrd.img
   map --unhook
   boot
   savedefault --wait=2

参考：

• https://wiki.grml.org/doku.php?id=rescueboot
• https://freedesktop.org/wiki/Software/systemd/Debugging/#index2h1
• https://forums.opensuse.org/t/shutdown-stalls-at-reached-target-shutdown-auto-power-off-occurs-4-minutes-later/138324
• https://github.com/tuxera/ntfs-3g/wiki/NTFS-3G-FAQ
• https://unix.stackexchange.com/questions/107978/cant-access-windows-drive-windows-is-hibernated-refused-to-mount
• https://askubuntu.com/questions/145902/unable-to-mount-windows-ntfs-filesystem-due-to-hibernation
• https://git.grml.org/?p=grml-live.git;a=blob_plain;f=templates/GRML/grml-cheatcodes.txt;hb=HEAD
• https://github.com/grml/grml/issues/206
• https://sources.debian.org/src/linux/3.16.56-1+deb8u1/Documentation/kernel-parameters.txt/
• https://manpages.debian.org/bookworm/live-boot-doc/live-boot.7.en.html
• http://bbs.wuyou.net/forum.php?mod=viewthread&tid=421620
• http://bbs.wuyou.net/forum.php?mod=viewthread&tid=208607

现象：一台机子，在 fdisk 里查看，没有分区。尝试使用 testdisk，盲猜是 gpt 分区类型，自动分析出有三个分区。写入、保存磁盘分区信息后，机子无法自动引导到操作系统，提示 1234F: 。

尝试：估计磁盘的原有分区是 mbr ，并不是 gpt ，因为里边没有 EFI 分区，且原系统中，没有 efi 相关的文件。然后再次运行 testdisk ，将分区信息保存为 mbr 。再次重启机子，依然是 1234F: ，无法引导系统。最后，在急救系统里，grub-install 到磁盘，重新安装 grub 。之后，就可以看到 grub 的引导信息，顺利进入到 debian 系统了。

参考：

• https://forum.proxmox.com/threads/boot-error.107097/
• https://pve.proxmox.com/wiki/Recover_From_Grub_Failure
• https://forum.cgsecurity.org/phpBB3/viewtopic.php?t=6635

尝试在debian 12.2系统里安装 iredmail-enterprise-1.0-beta1-linux-amd64。发现有两个问题：

1. web安装程序报错。搜索得，ntp 包对应的服务名是 ntpsec ，不是ntp 。 服务配置文件是：/lib/systemd/system/ntpsec.service
   TASK [ntp] Install ntp/chrony server to sync local time.
   ok: ntp

   TASK [ntp] Enable and restart ntp/chrony service.
   failed: ntp
   [ERROR] task error: failed in restarting service ntp: exit status 5, Failed to restart ntp.service: Unit ntp.service not found.

2. 通过web/admin 管理添加的用户，外网发信到这个用户会报错：said: 550 5.1.1 mailto:liu@iredmail.anqun.org: Recipient address rejected: User unknown (in reply to RCPT TO command) 这个用户是能登录到 roundcube。从服务器的shell里，也能通过命令行成功发信。作为对比，用web/iredadmin添加的用户，正常。删除之前报错的域，重新使用iredadmin添加回同一个域和用户，能从外网收信。尝试的解决方法是手动补上缺失的 forwardings 表内容：
   INSERT INTO forwardings (address, forwarding, domain, dest_domain, is_forwarding)
   VALUES ('liu@iredmail.anqun.org', 'liu@iredmail.anqun.org','iredmail.anqun.org', 'iredmail.anqun.org', 1);

参考：

• https://docs.iredmail.org/pro.html
• https://github.com/iredmail/iRedMail/blob/master/tools/create_mail_user_SQL.sh
• https://packages.debian.org/bookworm/ntp

问题：在 grml 的急救系统里无法启动 mariadb-server，报错：/usr/sbin/mariadbd $MYSQLD_OPTS $_WSREP_NEW_CLUSTER $_WSREP_START_POSITION (code=killed, signal=ABRT)

尝试：在 mysqld 的配置内容中，增加一行 innodb_flush_method = fsync

参考：

• https://jira.mariadb.org/browse/MDEV-28751
• https://groups.google.com/g/linux.debian.bugs.dist/c/q24pArKqbKo