rt/snapraid
1
0
Fork 0
mirror of https://github.com/amadvance/snapraid.git synced 2026-10-25 18:40:19 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Extend the -h option to check also copied/moved files

Browse source
This commit is contained in:
Andrea Mazzoleni 2015-09-29 22:31:13 +02:00
parent 254db2cdbf
commit 49bcad5448
6 changed files with 80 additions and 39 deletions
Download patch file Download diff file Expand all files Collapse all files

4
HISTORY
View file

@ -13,9 +13,11 @@ SnapRAID HISTORY
This allows to verify as early as possible that the written
parity during sync is really correct.
The "status" command reports the amount of scrub necessary to
ensure to scrub all the synched blocks.
ensure to scrub all the synced blocks.
* The -p, --percentage option supports the new the arguments:
"bad", "sync" and "full" to define the kind of scrub to do.
* Extends the -h, --pre-hash option to also verify moved and
copied files into the array before running a 'sync'.
* Update 'best' RAID functions for recent Atom CPUs.
* Validate filters specifications rejecting relative paths.

3
Makefile.am
View file

@ -257,7 +257,8 @@ endif
# Corrupt and move it to another disk
$(TESTENV) ./mktest$(EXEEXT) damage 1 1 1 bench/disk1/COPY
mv bench/disk1/COPY bench/disk2/COPY
# Now sync with failure as the data won't match
# Now sync with failure as the data won't match. We have two points of failure, pre-hash and sync
$(FAILENV) ./snapraid$(EXEEXT) $(CHECKFLAGS) -c $(CONF) --test-expect-failure -h sync
$(FAILENV) ./snapraid$(EXEEXT) $(CHECKFLAGS) -c $(CONF) --test-expect-failure sync
# Now sync with force-nocopy
$(TESTENV) ./snapraid$(EXEEXT) $(CHECKFLAGS) -c $(CONF) --force-nocopy sync

67
cmdline/sync.c
View file

@ -63,15 +63,15 @@ static int state_hash_process(struct snapraid_state* state, block_off_t blocksta
for (i = blockstart; i < blockmax; ++i) {
struct snapraid_block* block;
unsigned block_state;
block = disk_block_get(disk, i);
/* if no file in this block, skip it */
if (!block_has_file(block))
continue;
/* get the state of the block */
block_state = block_state_get(block);
/* if the block already has a hash, skip it */
if (block_has_updated_hash(block))
/* process REP and CHG blocks */
if (block_state != BLOCK_STATE_REP && block_state != BLOCK_STATE_CHG)
continue;
++countmax;
@ -99,20 +99,17 @@ static int state_hash_process(struct snapraid_state* state, block_off_t blocksta
struct snapraid_block* block;
int read_size;
unsigned char hash[HASH_SIZE];
unsigned block_state;
block = disk_block_get(disk, i);
/* if no file in this block, skip it */
if (!block_has_file(block))
continue;
/* get the state of the block */
block_state = block_state_get(block);
/* if the block already has a hash, skip it */
if (block_has_updated_hash(block))
/* process REP and CHG blocks */
if (block_state != BLOCK_STATE_REP && block_state != BLOCK_STATE_CHG)
continue;
/* here only CHG blocks are allowed */
assert(block_state_get(block) == BLOCK_STATE_CHG);
/* get block specific info */
info = info_get(&state->infoarr, i);
@ -256,14 +253,45 @@ static int state_hash_process(struct snapraid_state* state, block_off_t blocksta
memhash(state->hash, state->hashseed, hash, buffer, read_size);
}
/* copy the hash in the block */
memcpy(block->hash, hash, HASH_SIZE);
if (block_state == BLOCK_STATE_REP) {
/* compare the hash */
if (memcmp(hash, block->hash, HASH_SIZE) != 0) {
/* file we are processing for error reporting */
struct snapraid_file* file = block_file_get(block);
/* and mark the block as hashed */
block_state_set(block, BLOCK_STATE_REP);
log_tag("error:%u:%s:%s: Unexpected data change\n", i, disk->name, esc(file->sub));
log_error("Data change at file '%s' at position '%u'\n", handle[j].path, block_file_pos(block));
log_error("WARNING! Unexpected data modification of a file without parity!\n");
/* mark the state as needing write */
state->need_write = 1;
if (file_flag_has(file, FILE_IS_COPY)) {
log_error("This file was detected as a copy of another file with the same name, size,\n");
log_error("and timestamp, but the file data isn't matching the assumed copy.\n");
log_error("If this is a false positive, and the files are expected to be different,\n");
log_error("you can 'sync' anyway using 'snapraid --force-nocopy sync'\n");
} else {
log_error("Try removing the file from the array and rerun the 'sync' command!\n");
}
/* block sync to allow a recovery before overwriting */
/* the parity needed to make such recovery */
*skip_sync = 1; /* avoid to run the next sync */
++error;
continue;
}
} else {
/* the only other case is BLOCK_STATE_CHG */
assert(block_state == BLOCK_STATE_CHG);
/* copy the hash in the block */
memcpy(block->hash, hash, HASH_SIZE);
/* and mark the block as hashed */
block_state_set(block, BLOCK_STATE_REP);
/* mark the state as needing write */
state->need_write = 1;
}
/* count the number of processed block */
++countpos;
@ -1307,6 +1335,7 @@ int state_sync(struct snapraid_state* state, block_off_t blockstart, block_off_t
/* LCOV_EXCL_STOP */
}
/* save with the new hash information */
if (state->need_write)
state_write(state);
}

15
snapraid.1
View file

@ -1000,18 +1000,21 @@ option can speedup a lot the checking process.
This option can be used only with \[dq]check\[dq].
.TP
.B \-h, \-\-pre\-hash
In \[dq]sync\[dq] runs a preliminary hashing phase of all
the new data to verify the data used in the parity computation.
In \[dq]sync\[dq] runs a preliminary hashing phase of all the new data
to have an additional verification before the parity computation.
Usually in \[dq]sync\[dq] no preliminary hashing is done, and the new
data is hashed just before the parity computation when it\'s read
for the first time,
for the first time.
Unfortunately, this process happens when the system is under
heavy load, with all disks spinning and a busy CPU.
This is an extreme condition for your machine, and if it has a
heavy load, with all disks spinning and with a busy CPU.
This is an extreme condition for the machine, and if it has a
latent hardware problem, it\'s possible to encounter silent errors
what cannot be detected because the data is not yet hashed.
To avoid this risk, you can enable the \[dq]pre\-hash\[dq] mode and have
all the data hashed two times to ensure its integrity.
all the data read two times to ensure its integrity.
This option also verifies the files moved inside the array,
to ensure that the move operation went succesfully, and in case
to block the sync and to allow to run a fix operation.
This option can be used only with \[dq]sync\[dq].
.TP
.B \-i, \-\-import DIR

15
snapraid.d
View file

@ -714,18 +714,21 @@ Options
This option can be used only with "check".
-h, --pre-hash
In "sync" runs a preliminary hashing phase of all
the new data to verify the data used in the parity computation.
In "sync" runs a preliminary hashing phase of all the new data
to have an additional verification before the parity computation.
Usually in "sync" no preliminary hashing is done, and the new
data is hashed just before the parity computation when it's read
for the first time,
for the first time.
Unfortunately, this process happens when the system is under
heavy load, with all disks spinning and a busy CPU.
This is an extreme condition for your machine, and if it has a
heavy load, with all disks spinning and with a busy CPU.
This is an extreme condition for the machine, and if it has a
latent hardware problem, it's possible to encounter silent errors
what cannot be detected because the data is not yet hashed.
To avoid this risk, you can enable the "pre-hash" mode and have
all the data hashed two times to ensure its integrity.
all the data read two times to ensure its integrity.
This option also verifies the files moved inside the array,
to ensure that the move operation went succesfully, and in case
to block the sync and to allow to run a fix operation.
This option can be used only with "sync".
-i, --import DIR

15
snapraid.txt
View file

@ -773,18 +773,21 @@ SnapRAID provides the following options:
This option can be used only with "check".
-h, --pre-hash
In "sync" runs a preliminary hashing phase of all
the new data to verify the data used in the parity computation.
In "sync" runs a preliminary hashing phase of all the new data
to have an additional verification before the parity computation.
Usually in "sync" no preliminary hashing is done, and the new
data is hashed just before the parity computation when it's read
for the first time,
for the first time.
Unfortunately, this process happens when the system is under
heavy load, with all disks spinning and a busy CPU.
This is an extreme condition for your machine, and if it has a
heavy load, with all disks spinning and with a busy CPU.
This is an extreme condition for the machine, and if it has a
latent hardware problem, it's possible to encounter silent errors
what cannot be detected because the data is not yet hashed.
To avoid this risk, you can enable the "pre-hash" mode and have
all the data hashed two times to ensure its integrity.
all the data read two times to ensure its integrity.
This option also verifies the files moved inside the array,
to ensure that the move operation went succesfully, and in case
to block the sync and to allow to run a fix operation.
This option can be used only with "sync".
-i, --import DIR