journaling and file locking

Question 1

how are journaling file systems fast

Answer 1

it decouples the file writes from the disk head movement
the flush operation can do in-order block/sector writes
the transaction log could be on a faster disk/media

Question 2

how are journalling file systems resillient

Answer 2

the log can be replayed in event of system failure as transactins must be atomic and idempotent

Question 3

idempotent

Answer 3

can be repeated a number of times and still have the same result

Question 4

atomic

Answer 4

all or no subparts of the transaction must be complete

Question 5

how does a Journaling file system work

Answer 5

the file system updates are written as transactions to a journal or log
the transactions are periodically flushed to a disk
entries are only removed from the log when confirmed written/flushed

Question 6

how is a journaling file system circular

Answer 6

the log has a fixed size and reuses space by overwriting the oldest entries when its full

Question 7

shared lock

Answer 7

can have as many as you want
cannot include any data with an exclusive lock

Question 8

exclusive lock

Answer 8

only one transaction can hold the lock at a time
if a transaction requests a lock on a byte thats already locked then it fails

Question 9

blocking lock request

Answer 9

hangs until the bytes are available (have no other locks on them)

Question 10

non-blocking lock requests

Answer 10

immediately returns if the bytes are unavailable

Question 11

which operation are shared locks used for

Answer 11

reading

Question 12

which operation are exclusive locks used for

Answer 12

writing