M.I.T. DEPARTMENT OF EECS

6.033 - Computer System Engineering WAL Hands-On Assignment

Hands-on 7: Write Ahead Log System

Intro to wal-sys

Complete the following hands-on assignment. Do the activities described, and submit your solutions using the online submission site before 11:59p.

This hands-on assignment will give you some experience using a Write Ahead Log (WAL) system. This system corresponds to the WAL scheme described in Section 9.3 of the textbook. You should carefully read that section before attempting this assignment. You can do this hands-on on any computer that has a Python language interpreter, but we will be able to answer your questions more easily if you run this on an Athena workstation. You can download the WAL system from here (if your browser displays the file in a window instead of saving it, use "File -> Save As" to save the file). The downloaded file is a Python script named wal-sys. Before trying to run it, change its permissions to make it executable, for example by typing:

athena% chmod +x wal-sys.py

The wal-sys script can be run as follows:

athena% ./wal-sys.py [-reset]

Alternatively, you can run the script as:

athena% python ./wal-sys.py [-reset]

Wal-sys is a simple WAL system that models a bank's central database, implementing redo logging for error-recovery. Wal-sys creates and uses two files, named LOG and DB, in the current working directory. The "LOG" file contains the log entries, and the "DB" file contains all of the installed changes to the database.

After you start wal-sys, you can enter commands to manage recoverable actions and accounts. There are also commands to simulate a system crash and to print the contents of the "LOG" and "DB" files. All the commands to wal-sys are case sensitive. Since wal-sys uses the standard input stream, you can use the system in batch mode. To do this, place your commands in a file ("cmd.in" for example) and redirect the file to wal-sys's standard input:

athena% ./wal-sys.py -reset < cmd.in.

When using batch mode, make sure that each command is followed by a newline character (including the last one).

When you restart wal-sys, it will perform a log-based recovery of the "DB" file using the "LOG" file it finds in the current working directory. The -reset option tells wal-sys to discard the contents of any previous "DB" and "LOG" files so that it can start with a clean initial state.

Commands interpreted by wal-sys

The following commands are used for managing recoverable actions and accounts:

The following commands help us understand the dynamics of the WAL system:

Using wal-sys

Start wal-sys with a reset:
athena% ./wal-sys.py -reset
and run the following commands (sequence 1):
begin 1
create_account 1 studentA 1000
commit 1
end 1
begin 2
create_account 2 studentB 2000
begin 3
create_account 3 studentC 3000
credit_account 3 studentC 100
debit_account 3 studentA 100
commit 3
show_state
crash
Wal-sys should print out the contents of the "DB" and "LOG" files, and then exit.

Use a text editor to examine the "DB" and "LOG" files and answer the following questions (do not run wal-sys again until you have answered these questions):

Question 1: Wal-sys displays the current state of the database contents after you type show_state. Why doesn't the database show studentB?

Question 2: When the database recovers, which accounts should be active, and what values should they contain?

Question 3: Can you explain why the "DB" file does not contain a record for studentC and contains the pre-debit balance for studentA?

Recovering the database

When you run wal-sys without the -reset option it recovers the database "DB" using the "LOG" file. To recover the database and then look at the results, type:
athena% ./wal-sys.py
> show_state
> crash

Question 4: What do you expect the state of "DB" to be after wal-sys recovers after this new crash? Why?

Question 5: Run wal-sys again to recover the database. Examine the "DB" file. Does the state of the database match your expectations? Why or why not?

Question 6: During recovery, wal-sys reports the action_ids of those recoverable actions that are "Losers", "Winners", and "Done". What is the meaning of these categories?

Checkpoints

Start wal-sys with a reset:
athena% ./wal-sys.py -reset
and run the following commands (sequence 2):
begin 1
create_account 1 studentA 1000
commit 1
end 1
begin 2
create_account 2 studentB 2000
checkpoint
begin 3
create_account 3 studentC 3000
credit_account 3 studentC 100
debit_account 2 studentB 100
commit 3
show_state
crash
Note: the remainder of this assignment is only concerned with sequence 2. We will ask you to crash and recover the system a few times, but you should not run the sequence commands again. (Also note that in sequence 2, the command debit_account 2 studentB 100 refers to action_id 2, not action_id 3! This is not a typo).

Question 7: Why are the results of recoverable action 2's create_account 2 studentB 2000 command not installed in "DB" by the checkpoint command on the following line?

Examine the "LOG" output file. In particular, inspect the CHECKPOINT entry. Also, count the number of entries in the "LOG" file. Run wal-sys again to recover the database.

Question 8: How many lines were rolled back? What is the advantage of using checkpoints?

Note down the action_ids of "Winners", "Losers", and "Done". Use the show_state command to look at the recovered database and verify that the database recovered correctly. Crash the system, and then run wal-sys again to recover the database a second time.

Question 9: Does the second run of the recovery procedure (for sequence 2) restore "DB" to the same state as the first run? What is this property called?

Question 10: Compare the action_ids of "Winners", "Losers", and "Done" from the second recovery with those from the first. The lists are different. How does the recovery procedure guarantee the property from Question 9 even though the recovery procedure can change? (Hint: Examine the "LOG" file).

Optional: Wal-sys has a hitherto unmentioned option: if you type wal-sys -undo it will perform undo logging and undo recovery. Try the above sequences again with undo logging to see what changes.

Question 11: How long did this hands-on take you?



Go to 6.033 Home Page