Booting and Shutting Down

1. Booting and Shutting Down Chapter 2

2. Introduction • UNIX is a complex operating system, and turning UNIX systems on and off is more complicated than just flipping the power switch. • Both operations must be performed correctly if the system is to stay healthy. Chapter 2 - Booting and Shutting Down

3. 1. Bootstrapping • Introduction: • Bootstrapping is the nerd word for “starting up a computer” • During this process, the kernel is loaded into memory and begins to execute. A variety of initialization tasks are performed, and the system is made available to users. Chapter 2 - Booting and Shutting Down

4. 1. Bootstrapping • Boot time is a period of special vulnerability. All of the following can prevent a computer from coming up • Errors in configuration files • Missing or unreliable equipment • Damaged filesystems • Dealing with Boot configuration is often one of the first tasks an administer must perform on a new system. • Unfortunately it is also one of the most difficult Chapter 2 - Booting and Shutting Down

5. 1. Bootstrapping • The Boot Process • Power on – • prom code is executed, then how do I load and start the kernel? • Kernel loaded – • Hardware probed, then init is spawned (PID 1) • Init (all controlled by rc files) • File systems checked and mounted • System demons started Chapter 2 - Booting and Shutting Down

6. 1. Bootstrapping • Automatic and manual booting • Most UNIX systems can boot in either automatic mode or manual mode. • In automatic, the system performs the complete boot procedure on its own. • In manual mode, the system follows the automatic procedure up to a point and then turns control over to the operator before most init scripts have run. At this point the machine is in single-user mode. Chapter 2 - Booting and Shutting Down

7. 1. Bootstrapping • You usually have to boot manually when some problems break automatic booting • corrupt file system, • improperly configured network device • Therefore it is important to understand the process of how to perform a manual boot. Chapter 2 - Booting and Shutting Down

8. 1. Bootstrapping • Steps in the boot process • Loading and initialization of the kernel • Device detection and configuration • Creation of spontaneous system processes • Operator intervention (manual boot) • Execution of system startup scripts • Multi-user operation • We effect most bootstrap configuration by editing the system startup scripts Chapter 2 - Booting and Shutting Down

9. 1. Bootstrapping • Kernel initialization • The first bootstrapping task is to get this program into memory so that it can be executed • Most systems have a two stage process • The PROM loads a small boot program into memory from disk. • This program arranges for the kernel to be loaded • The kernel then find out how much memory is available • Many of the kernel’s internal data structures are statically sized, so the kernel sets aside some memory for itself. Chapter 2 - Booting and Shutting Down

10. 1. Bootstrapping • Hardware configuration • One of the first chores is to check out the environment to see what hardware is present. • When you construct a kernel for your system, you tell it what hardware devices it should expect to find • When the kernel begins to execute, it tries to locate and initialize each device that you have told it about. Chapter 2 - Booting and Shutting Down

11. 1. Bootstrapping • System processes • Once basic initialization is complete, the kernel creates several “spontaneous” processes in user space. • They are spontaneous because they are not created by the normal UNIX fork mechanism. • On a System V-ish machine they are: • sched – process 0 • init – process 1 • Then various memory and kernel process handlers Chapter 2 - Booting and Shutting Down

12. 1. Bootstrapping • Of all these processes, only init is really a full-fledged user process. • The others are actually portions of the kernel that have been dressed up to look like processes for scheduling or architectural reasons. • Now init has to create the daemons and the processes to handle basic operations (like accepting logins) Chapter 2 - Booting and Shutting Down

13. 1. Bootstrapping • Operator intervention (manual boot only) • Single-User Mode • Usually prompted for the root password • Then given a shell • One many systems /root is mounted read-only • if /tmp is part of /root then some programs will not work properly (vi, …) • Most other filesystems are not mounted. • This is good if you have a sick filesystem. Chapter 2 - Booting and Shutting Down

14. 1. Bootstrapping • Execution of startup scripts • At this point there are no more magic steps in the boot process. • The startup scripts are just normal shell scripts and they are selected and run by init according to an algorithm that, though sometimes tortuous, is eminently comprehensible. • The care, feeding, and taxonomy of startup scripts merits a major section of its own and is taken up in Section 4. Chapter 2 - Booting and Shutting Down

15. 1. Bootstrapping • Multi-user operation • After the initialization scripts have run, the system is fully operational • except no one can log in. • For logins to occur a getty process must be listening to a particular terminal • init spawns several getty processes • init is also responsible for spawning graphical login systems such as xdm, gdm, or dtlogin. Chapter 2 - Booting and Shutting Down

16. 2. Booting PCs • At this point we have seen the general outline of the boot process. We are now going to go back and fill in some of the details. • We will start with the initial power-on sequence and the loading of the kernel. • On traditional hardware this is a straightforward process • PC booting, on the other hand, is a lengthy ordeal that requires a bit of background information to understand. Chapter 2 - Booting and Shutting Down

17. 2. Booting PCs • How a PC is different from proprietary hardware • When a machine boots, it begins by executing code stored in ROMs. • On a machine designed explicitly for UNIX, the code typically knows how to use devices connected to the machine, how to talk to the network on a basic level, and how to understand disk-based filesystems. • This is convenient because you can just type the name of a new kernel and the firmware will know how to locate and read that file. Chapter 2 - Booting and Shutting Down

18. 2. Booting PCs • On PC’s the initial boot code is called a BIOS • It is extremely simplistic compared to the firmware of a UNIX machine. • Actually a PC has several level of BIOS • one for the machine itself, • One for the VIDEO card • one for the SCSI card (if present) • … • Working out the conflicts and interactions among the various BIOSes can be a real nightmare. Chapter 2 - Booting and Shutting Down

19. 2. Booting PCs • The PC boot process • The BIOS will normally let you select which devices you want to boot from. • Once your machine has figured out what device to boot from, it will try to load the first 512 bytes of the disk (the Master Boot Record - MBR) • The MBR tells the computer from which partition to load a secondary program (the boot loader) • More info is given in Chapter 8 • The default MBR just loads from the first partition • more sophisticated ones now allow multiple OS’s (and kernels) • The boot loader is responsible for loading the kernel Chapter 2 - Booting and Shutting Down

20. 2. Booting PCs • Examples of boot loaders include • LILO: the Linux boot loader • The FreeBSD boot loader • … • Your text covers how to use and configure these in detail. Chapter 2 - Booting and Shutting Down

21. 2. Booting PCs • Multibooting on PCs • Since many operating systems run on PCs, it is fairly common to set up a machine to be able to boot several different systems. • To make this work you need to configure a boot loader to recognize all of the different operating systems on your disk. Chapter 2 - Booting and Shutting Down

22. 3. Booting in Single-User Mode • This section gives the nitty-gritty details of single-user booting on each of our example operating systems. • Sloaris single-user • HP-UX single-user • Linux single-user • FreeBSD single-user Chapter 2 - Booting and Shutting Down

23. 3. Booting in Single-User Mode • Notice that the options given in LILO a Linux Boot Loader are similar to the options given in the Sun Boot Prom. • Except that you can do certain hardware things from the SUM firmware that you can’t in a boot loader • like probe the SCSI devices Chapter 2 - Booting and Shutting Down

24. 4. Startup Scripts • Introduction • After you exit single-user mode, init executes the system startup scripts • System scripts can be organized in two principal ways, each deeply rooted in ancient history. • On systems descended from BSD, the scripts are kept in /etc and have names starting with the letters rc. • On System-V derived systems, the scripts are kept in /etc/init.d and links to them are made in the directories /etc/rc0.d, /etc/rc1.d,... Chapter 2 - Booting and Shutting Down

25. 4. Startup Scripts • Some tasks that are often performed by the startup scripts • Setting the name of the computer • Setting the time zone • Checking the disks with fsck (in automatic mode) • Mounting the systems disks • Removing the old files from /tmp • Configuring the network interfaces • Starting up daemons and network services Chapter 2 - Booting and Shutting Down

26. 4. Startup Scripts • Most startup scripts are quite verbose and print out a description of everything they are doing. • On systems of yore, it was common practice to modify startup scripts to make them do the right thing for a particular environment. • These days, you put the details in a separate configuration file that the scripts consult. • They generally set the values of certain shell variables. Chapter 2 - Booting and Shutting Down

27. 4. Startup Scripts • System V-style startup scripts • These are the most common today. • The System V init defines 7 “run levels” • Level 0 - the system is completely shut down. • Level 1 - single user mode • Level 2 - 5 are multi-user levels • Level 6 is a “reboot” level Chapter 2 - Booting and Shutting Down

28. 4. Startup Scripts • Level 0 and 6 are special • The system can’t actually remain in them; it shuts down or reboots a s a side effect of entering them. • On most systems, the normal multi-user run level is 2 or 3 (4 and 5 are rarely used) • There seem to be more run levels than are necessary or useful • The usual explanation is that a phone switch had 7 run levels, so a UNIX system should have at least that many • Red Hat actually supports 10 (7 through 9 are undefined) Chapter 2 - Booting and Shutting Down

29. 4. Startup Scripts • /etc/inittab tells init what to do at each of its run levels • The format varies from system to system, but the basic idea is that inittab defines commands that are to be run (or kept running) when the system enters each level. • It is usually not necessary for system administrators to deal directly with /etc/inittab because the script-based interface is adequate for almost any application Chapter 2 - Booting and Shutting Down

30. 4. Startup Scripts • The master copies of the startup scripts live in a directory called init.d (usually in /etc) • Each script is responsible for one daemon or one particular aspect of the system. • The scripts understand the arguments start and stop. • Most also understand restart. • When moving from one level to another it looks in rclevel.d (rc0.d, rc1.d,…) Chapter 2 - Booting and Shutting Down

31. 4. Startup Scripts • This directory (rclevel.d) typically contains symbolic links that point back to init.d • The names start with S or K followed by a number and name for example: S34named • The S and K tell init.d what to start or kill (on the way up or down) • and what order to do it in • ascending on the way up, • descending on the way down Chapter 2 - Booting and Shutting Down

32. 4. Startup Scripts • Solaris startup scripts • Solaris, HP-UX, and Red Hat all use System V-style startup scripts that are kept in an init.d directory. • Under Solaris, both the init.d and rclevel.d directories are in /etc. Chapter 2 - Booting and Shutting Down

33. 4. Startup Scripts • HP-UX startup scripts • Under HP-UX, the actual startup scripts are kept in /sbin/init.d • The run-level directories are also in /sbin • Config files that change the behavior of the startup scripts generally live in /etc/rc.config.d • Their names correspond to the names of the startup scripts in /sbin/init.d • Output is put n /etc/rc.log Chapter 2 - Booting and Shutting Down

34. 4. Startup Scripts • Red Hat startup scripts • Startup scripts are the one thing that distinguish Linux distributions from each other. • Debian uses startup scripts similar to Solaris • Slackware uses scripts that are similar to FreeBSD’s • Red Hat uses a hybrid of the System V and BSD systems, with a few twists thrown in just to make life difficult for everyone. Chapter 2 - Booting and Shutting Down

35. 4. Startup Scripts • FreeBSD startup scripts • FreeBSD’s init runs only a single startup script /etc/rc • This master script then runs the system’s other rc scripts • all of which live in /etc and are called rc.something Chapter 2 - Booting and Shutting Down

36. 5. Rebooting and shutting down • UNIX file systems buffer changes in memory and write them back to disk only sporadically. • This scheme makes disk I/O faster, but it also makes the file systems susceptible to data loss when the system is rudely halted • It is always a god idea to shut down a machine nicely when possible. • Improper shutdowns can result in anything from subtle, insidious problems to major catastrophe. Chapter 2 - Booting and Shutting Down

37. 5. Rebooting and shutting down • On non-UNIX operating systems, rebooting the operating system is an appropriate first course of treatment for almost any problem. • On a UNIX system, it is better to think first and reboot second. • You usually need to reboot when you add a new piece of hardware or when existing hardware becomes so confused that it cannot be reset. Chapter 2 - Booting and Shutting Down

38. 5. Rebooting and shutting down • Unlike bootstrapping, which can be done in essentially only one way, there are a number of ways to shut down or reboot. They are: • Turning off the power. • Using the shutdown command • Using the halt and reboot commands • Sending init a TERM signal • Using telinit to change init’s run level • Killing init. Chapter 2 - Booting and Shutting Down

39. 5. Rebooting and shutting down • Turning off the power • Even on small UNIX systems, turning off the power is not a good way to shut down • You can potentially loose data and leave the system’s files in an inconsistent state. • Some machines have a “soft power switch” • when you press the button, the machine actually runs some commands to perform a proper shutdown sequence. Chapter 2 - Booting and Shutting Down

40. 5. Rebooting and shutting down • shutdown: the genteel way to halt the system • shutdown is the safest, most considerate, and most thorough way to initiate and halt or reboot or return to single-user mode. • Unfortunately, almost every vendor has decided to tamper with its arguments. Chapter 2 - Booting and Shutting Down

41. 5. Rebooting and shutting down • You can ask shutdown to wait a while before bringing down the system. • During that waiting period, shutdown sends messages to logged-in users at progressively shorter intervals, warning them of the impending downtime. • You can also supply a short message of your own • Your message should tell why the system is being brought down and should estimate how long it will be before users can log in again. Chapter 2 - Booting and Shutting Down

42. 5. Rebooting and shutting down • halt: a simpler way to shut down • The halt command performs the essential duties required to bring the system down. • It is called by shutdown -h • halt • logs the shutdown, • kills the nonessential processes, • executes the sync system call, • waits for the filesystem writes to complete, • and then halts the kernel Chapter 2 - Booting and Shutting Down

43. 5. Rebooting and shutting down • reboot: quick and dirty restart • reboot is almost identical to halt, but it causes the machine to reboot from scratch rather than halting. • reboot is called by shutdown -r • Like halt it supports the -n and -q flags Chapter 2 - Booting and Shutting Down

44. 5. Rebooting and shutting down • Sending init a TERM signal • The results of killing init are unpredictable and often nasty. • When the BSD version of init receives a TERM signal, it usually kills all user processes, daemons, and gettys and returns the system to single user mode. • This facility is used by shutdown • More information about signals is given in Chapter 4. Chapter 2 - Booting and Shutting Down

45. 5. Rebooting and shutting down • telinit: change init’s run level • On systems with the beefy multilevel init, you can use the telinit command to direct init to go to a specific run level • telinit is most useful for testing changes to the inittab file Chapter 2 - Booting and Shutting Down

46. 5. Rebooting and shutting down • Killing init • init is so important to the operation of the system that if it is killed with kill -KILL or kill -9, most computers will reboot automatically (some kernels just panic) • This is a rough way to reboot • Use shutdown or reboot instead Chapter 2 - Booting and Shutting Down

47. Chapter 2 - Booting and Shutting Down