comment: <> (SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT) comment: <> (LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,) comment: <> (DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY) comment: <> (THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT) comment: <> ((INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE) comment: <> (OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.)

NAME
SYNOPSIS
DESCRIPTION
POOL CHECKING FUNCTIONS
POOLSET SYNCHRONIZATION AND TRANSFORMATION
LIBRARY API VERSIONING
DEBUGGING AND ERROR HANDLING
EXAMPLE
ACKNOWLEDGEMENTS
SEE ALSO

NAME

libpmempool – pool management library

SYNOPSIS

#include <libpmempool.h>
cc -std=gnu99 ... -lpmempool -lpmem

Health check functions:

PMEMpoolcheck *pmempool_check_init(struct pmempool_check_args *args,
	size_t args_size);
struct pmempool_check_status *pmempool_check(PMEMpoolcheck *ppc);
enum pmempool_check_result pmempool_check_end(PMEMpoolcheck *ppc);

Poolset synchronization and transformation:

int pmempool_sync(const char *poolset_file, unsigned flags); (EXPERIMENTAL)

int pmempool_transform(const char *poolset_file_src,
	const char *poolset_file_dst,
	unsigned flags); (EXPERIMENTAL)

Library API versioning:

const char *pmempool_check_version(unsigned major_required,
	unsigned minor_required);

Error handling:

const char *pmempool_errormsg(void);

DESCRIPTION

libpmempool provides a set of utilities for off-line analysis and manipulation of a pool. By pool in this manpage we mean pmemobj pool, pmemblk pool, pmemlog pool or BTT layout, independent of the underlying storage. Some of libpmempool functions are required to work without any impact on processed pool but some of them may create a new or modify an existing one.

libpmempool is for applications that need high reliability or built-in troubleshooting. It may be useful for testing and debugging purposes also.

POOL CHECKING FUNCTIONS

To perform check provided by libpmempool, a check context must be first initialized using pmempool_check_init() function described in this section. Once initialized check context is represented by an opaque handle, of type PMEMpoolcheck*, which is passed to all of the other functions described in this section.

To execute check pmempool_check() must be called iteratively. Each call resumes check till new status will be generated. Each status is represented by struct pmempool_check_status* structure. It may carry various types of messages described in this section.

When check is completed pmempool_check() returns NULL pointer. Check must be finalized using pmempool_check_end(). It returns enum pmempool_check_result describing result of the whole check.

NOTE

Currently, checking the consistency of a pmemobj pool is not supported.

PMEMpoolcheck *pmempool_check_init(struct pmempool_check_args *args,
	size_t args_size);

The pmempool_check_init() initializes check context. args describes parameters of the check context. args_size should be equal to the size of the struct pmempool_check_args. struct pmempool_check_args is defined as follows:

struct pmempool_check_args
{
	/* path to the pool to check */
	const char *path;

	/* optional backup path */
	const char *backup_path;

	/* type of the pool */
	enum pmempool_pool_type pool_type;

	/* parameters */
	int flags;
};

The flags argument accepts any combination of the following values (ORed):

• PMEMPOOL_CHECK_REPAIR - perform repairs
• PMEMPOOL_CHECK_DRY_RUN - emulate repairs, not supported on Device DAX
• PMEMPOOL_CHECK_ADVANCED - perform hazardous repairs
• PMEMPOOL_CHECK_ALWAYS_YES - do not ask before repairs
• PMEMPOOL_CHECK_VERBOSE - generate info statuses
• PMEMPOOL_CHECK_FORMAT_STR - generate string format statuses

If provided parameters are invalid or initialization process fails pmempool_check_init() returns NULL and sets errno appropriately. pool_type has to match type of the pool being processed. You can turn on pool type detection by setting pool_type to PMEMPOOL_POOL_TYPE_DETECT. Pool type detection fail ends check.

backup_path argument can either be:

• NULL. It indicates no backup will be performed.

• a non existing file. It is valid only in case path is a single file pool. It indicates a backup_path file will be created and backup will be performed.

• an existing poolset file of the same structure (the same number of parts with exactly the same size) as the source poolset. It is valid only in case path is a poolset. It indicates backup will be performed in a form described by the backup_path poolset.

Backup is supported only if the source poolset has no defined replicas.

Poolsets with remote replicas are not supported neither as path nor as backup_path.

This is an example of a check context initialization:

struct pmempool_check_args args =
{
	.path = "/path/to/blk.pool",
	.backup_path = NULL,
	.pool_type = PMEMPOOL_POOL_TYPE_BLK,
	.flags = PMEMPOOL_CHECK_REPAIR | PMEMPOOL_CHECK_DRY_RUN |
		PMEMPOOL_CHECK_VERBOSE | PMEMPOOL_CHECK_FORMAT_STR
};

PMEMpoolcheck *ppc = pmempool_check_init(&args, sizeof(args));

The check will process a pool of type PMEMPOOL_POOL_TYPE_BLK located in the path /path/to/blk.pool. Before check it will not create a backup of the pool (backup_path == NULL). If the check will find any issues it will try to perform repair steps (PMEMPOOL_CHECK_REPAIR), but it will not make any changes to the pool (PMEMPOOL_CHECK_DRY_RUN) and it will not perform any dangerous repair steps (no PMEMPOOL_CHECK_ADVANCED). The check will ask before performing any repair steps (no PMEMPOOL_CHECK_ALWAYS_YES). It will also generate detailed information about the check (PMEMPOOL_CHECK_VERBOSE). PMEMPOOL_CHECK_FORMAT_STR flag indicates string format statuses (struct pmempool_check_status). Currently it is the only supported status format so this flag is required.

struct pmempool_check_status *pmempool_check(PMEMpoolcheck *ppc);

The pmempool_check() function starts or resumes the check indicated by ppc. When next status will be generated it pauses the check and returns a pointer to the struct pmempool_check_status structure:

struct pmempool_check_status
{
	enum pmempool_check_msg_type type; /* type of the status */
	struct
	{
		const char *msg; /* status message string */
		const char *answer; /* answer to message if applicable */
	} str;
};

This structure can describe three types of statuses:

• PMEMPOOL_CHECK_MSG_TYPE_INFO - detailed information about the check. Generated only if a PMEMPOOL_CHECK_VERBOSE flag was set.
• PMEMPOOL_CHECK_MSG_TYPE_ERROR - encountered error
• PMEMPOOL_CHECK_MSG_TYPE_QUESTION - question. Generated only if an PMEMPOOL_CHECK_ALWAYS_YES flag was not set. It requires answer to be set to “yes” or “no” before continuing.

After calling pmempool_check() again the previously provided struct pmempool_check_status* pointer must be considered invalid. When the check completes pmempool_check() returns NULL pointer.

enum pmempool_check_result pmempool_check_end(PMEMpoolcheck* ppc);

The pmempool_check_end() function finalizes the check and releases all related resources. ppc is not a valid pointer after calling pmempool_check_end(). It returns enum pmempool_check_result summarizing result of the finalized check. pmempool_check_end() can return one of the following values:

• PMEMPOOL_CHECK_RESULT_CONSISTENT - the pool is consistent
• PMEMPOOL_CHECK_RESULT_NOT_CONSISTENT - the pool is not consistent
• PMEMPOOL_CHECK_RESULT_REPAIRED - the pool has issues but all repair steps completed successfully
• PMEMPOOL_CHECK_RESULT_CANNOT_REPAIR - the pool has issues which can not be repaired
• PMEMPOOL_CHECK_RESULT_ERROR - the pool has errors or the check encountered issue

POOLSET SYNCHRONIZATION AND TRANSFORMATION

POOLSET SYNC

int pmempool_sync(const char *poolset_file, unsigned flags); (EXPERIMENTAL)

The pmempool_sync() function synchronizes data between replicas within a poolset.

pmempool_sync() accepts two arguments:

• poolset_file - a path to a poolset file,

• flags - a combination of flags (ORed) which modify the way of synchronization.

NOTE: Only the poolset file used to create the pool should be used for syncing the pool.

The following flags are available:

• PMEMPOOL_DRY_RUN - do not apply changes, only check for viability of synchronization.

pmempool_sync() function checks if metadata of all replicas in a poolset are consistent, i.e. all parts are healthy, and if any of them is not, the corrupted or missing parts are recreated and filled with data from one of the healthy replicas.

The function returns either 0 on success or -1 in case of error with proper errno set accordingly.

NOTE: The pmempool_sync() API is experimental and it may change in future versions of the library.

POOLSET TRANSFORM

int pmempool_transform(const char *poolset_file_src,
	const char *poolset_file_dst,
	unsigned flags); (EXPERIMENTAL)

The pmempool_transform() function modifies internal structure of a poolset. It supports the following operations:

• adding one or more replicas,

• removing one or more replicas,

• reordering of replicas.

Currently these operations are allowed only for pmemobj pools (see libpmemobj(3)).

pmempool_transform() accepts three arguments:

• poolset_file_src - a path to a poolset file which defines the source poolset to be changed,

• poolset_file_dst - a path to a poolset file which defines the target structure of the poolset,

• flags - a combination of flags (ORed) which modify the way of synchronization.

The following flags are available:

• PMEMPOOL_DRY_RUN - do not apply changes, only check for viability of synchronization.

When adding or deleting replicas, the two poolset files can differ only in the definitions of replicas which are to be added or deleted. One cannot add and remove replicas in the same step. Only one of these operations can be performed at a time. Reordering replicas can be combined with any of them. Also, to add a replica it is necessary for its effective size to match or exceed the pool size. Otherwise the whole operation fails and no changes are applied. Effective size of a replica is the sum of sizes of all its part files decreased by 4096 bytes per each part file. The 4096 bytes of each part file is utilized for storing internal metadata of the pool part files.

The function returns either 0 on success or -1 in case of error with proper errno set accordingly.

NOTE: The pmempool_transform() API is experimental and it may change in future versions of the library.

LIBRARY API VERSIONING

This section describes how the library API is versioned, allowing applications to work with an evolving API.

const char *pmempool_check_version(
	unsigned major_required,
	unsigned minor_required);

The pmempool_check_version() function is used to see if the installed libpmempool supports the version of the library API required by an application. The easiest way to do this for the application is to supply the compile-time version information, supplied by defines in <libpmempool.h>, like this:

reason = pmempool_check_version(PMEMPOOL_MAJOR_VERSION,
                                PMEMPOOL_MINOR_VERSION);
if (reason != NULL) {
	/* version check failed, reason string tells you why */
}

Any mismatch in the major version number is considered a failure, but a library with a newer minor version number will pass this check since increasing minor versions imply backwards compatibility.

An application can also check specifically for the existence of an interface by checking for the version where that interface was introduced. These versions are documented in this man page as follows: unless otherwise specified, all interfaces described here are available in version 1.0 of the library. Interfaces added after version 1.0 will contain the text introduced in version x.y in the section of this manual describing the feature.

When the version check performed by pmempool_check_version() is successful, the return value is NULL. Otherwise the return value is a static string describing the reason for failing the version check. The string returned by pmempool_check_version() must not be modified or freed.

DEBUGGING AND ERROR HANDLING

Two versions of libpmempool are typically available on a development system. The normal version, accessed when a program is linked using the -lpmempool option, is optimized for performance. That version skips checks that impact performance and exceptionally logs any trace information or performs any run-time assertions. If an error is detected during the call to libpmempool function, an application may retrieve an error message describing the reason of failure using the following function:

const char *pmempool_errormsg(void);

The pmempool_errormsg() function returns a pointer to a static buffer containing the last error message logged for current thread. The error message may include description of the corresponding error code (if errno was set), as returned by strerror(3). The error message buffer is thread-local; errors encountered in one thread do not affect its value in other threads. The buffer is never cleared by any library function; its content is significant only when the return value of the immediately preceding call to libpmempool function indicated an error, or if errno was set. The application must not modify or free the error message string, but it may be modified by subsequent calls to other library functions.

A second version of libpmempool, accessed when a program uses the libraries under /usr/lib/nvml_debug, contains run-time assertions and trace points. The typical way to access the debug version is to set the environment variable LD_LIBRARY_PATH to /usr/lib/nvml_debug or /usr/lib64/nvml_debug depending on where the debug libraries are installed on the system. The trace points in the debug version of the library are enabled using the environment variable PMEMPOOL_LOG_LEVEL, which can be set to the following values:

• 0 - This is the default level when PMEMPOOL_LOG_LEVEL is not set. No log messages are emitted at this level.

• 1 - Additional details on any errors detected are logged (in addition to returning the errno-based errors as usual). The same information may be retrieved using pmempool_errormsg().

• 2 - A trace of basic operations is logged.

• 3 - This level enables a very verbose amount of function call tracing in the library.

• 4 - This level enables voluminous and fairly obscure tracing information that is likely only useful to the libpmempool developers.

The environment variable PMEMPOOL_LOG_FILE specifies a file name where all logging information should be written. If the last character in the name is “-”, the PID of the current process will be appended to the file name when the log file is created. If PMEMPOOL_LOG_FILE is not set, the logging output goes to stderr.

Setting the environment variable PMEMPOOL_LOG_FILE has no effect on the non-debug version of libpmempool.

EXAMPLE

The following example illustrates how the libpmempool API is used. The program detects the type and checks consistency of given pool. If there are any issues detected, the pool is automatically repaired.

#include <stddef.h>
#include <unistd.h>
#include <stdlib.h>
#include <stdio.h>
#include <libpmempool.h>

#define PATH "./pmem-fs/myfile"
#define CHECK_FLAGS (PMEMPOOL_CHECK_FORMAT_STR|PMEMPOOL_CHECK_REPAIR|\
                     PMEMPOOL_CHECK_VERBOSE)

int
main(int argc, char *argv[])
{
	PMEMpoolcheck *ppc;
	struct pmempool_check_status *status;
	enum pmempool_check_result ret;

	/* arguments for check */
	struct pmempool_check_args args = {
		.path		= PATH,
		.backup_path	= NULL,
		.pool_type	= PMEMPOOL_POOL_TYPE_DETECT,
		.flags		= CHECK_FLAGS
	};

	/* initialize check context */
	if ((ppc = pmempool_check_init(&args, sizeof(args))) == NULL) {
		perror("pmempool_check_init");
		exit(EXIT_FAILURE);
	}

	/* perform check and repair, answer 'yes' for each question */
	while ((status = pmempool_check(ppc)) != NULL) {
		switch (status->type) {
		case PMEMPOOL_CHECK_MSG_TYPE_ERROR:
			printf("%s\n", status->str.msg);
			break;
		case PMEMPOOL_CHECK_MSG_TYPE_INFO:
			printf("%s\n", status->str.msg);
			break;
		case PMEMPOOL_CHECK_MSG_TYPE_QUESTION:
			printf("%s\n", status->str.msg);
			status->str.answer = "yes";
			break;
		default:
			pmempool_check_end(ppc);
			exit(EXIT_FAILURE);
		}
	}

	/* finalize the check and get the result */
	ret = pmempool_check_end(ppc);
	switch (ret) {
		case PMEMPOOL_CHECK_RESULT_CONSISTENT:
		case PMEMPOOL_CHECK_RESULT_REPAIRED:
			return 0;
		default:
			return 1;
	}
}

See http://pmem.io/nvml/libpmempool for more examples using the libpmempool API.

ACKNOWLEDGEMENTS

libpmempool builds on the persistent memory programming model recommended by the SNIA NVM Programming Technical Work Group: http://snia.org/nvmp

SEE ALSO

mmap(2), munmap(2), msync(2), strerror(3), libpmemobj(3), libpmemblk(3), libpmemlog(3), libpmem(3) and http://pmem.io