28.5. Dynamic Tracing

PostgreSQL provides facilities to support dynamic tracing of the database server. This allows an external utility to be called at specific points in the code and thereby trace execution.

A number of probes or trace points are already inserted into the source code. These probes are intended to be used by database developers and administrators. By default the probes are not compiled into PostgreSQL; the user needs to explicitly tell the configure script to make the probes available.

Currently, the DTrace utility is supported, which, at the time of this writing, is available on Solaris, macOS, FreeBSD, NetBSD, and Oracle Linux. The SystemTap project for Linux provides a DTrace equivalent and can also be used. Supporting other dynamic tracing utilities is theoretically possible by changing the definitions for the macros in src/include/utils/probes.h.

28.5.1. Compiling for Dynamic Tracing

By default, probes are not available, so you will need to explicitly tell the configure script to make the probes available in PostgreSQL. To include DTrace support specify --enable-dtrace to configure. See Section 17.4 for further information.

28.5.2. Built-in Probes

A number of standard probes are provided in the source code, as shown in Table 28.47; Table 28.48 shows the types used in the probes. More probes can certainly be added to enhance PostgreSQL's observability.

Table 28.47. Built-in DTrace Probes

NameParametersDescription

transaction-start

(LocalTransactionId)

Probe that fires at the start of a new transaction. arg0 is the transaction ID.

transaction-commit

(LocalTransactionId)

Probe that fires when a transaction completes successfully. arg0 is the transaction ID.

transaction-abort

(LocalTransactionId)

Probe that fires when a transaction completes unsuccessfully. arg0 is the transaction ID.

query-start

(const char *)

Probe that fires when the processing of a query is started. arg0 is the query string.

query-done

(const char *)

Probe that fires when the processing of a query is complete. arg0 is the query string.

query-parse-start

(const char *)

Probe that fires when the parsing of a query is started. arg0 is the query string.

query-parse-done

(const char *)

Probe that fires when the parsing of a query is complete. arg0 is the query string.

query-rewrite-start

(const char *)

Probe that fires when the rewriting of a query is started. arg0 is the query string.

query-rewrite-done

(const char *)

Probe that fires when the rewriting of a query is complete. arg0 is the query string.

query-plan-start

()

Probe that fires when the planning of a query is started.

query-plan-done

()

Probe that fires when the planning of a query is complete.

query-execute-start

()

Probe that fires when the execution of a query is started.

query-execute-done

()

Probe that fires when the execution of a query is complete.

statement-status

(const char *)

Probe that fires anytime the server process updates its pg_stat_activity.status. arg0 is the new status string.

checkpoint-start

(int)

Probe that fires when a checkpoint is started. arg0 holds the bitwise flags used to distinguish different checkpoint types, such as shutdown, immediate or force.

checkpoint-done

(int, int, int, int, int)

Probe that fires when a checkpoint is complete. (The probes listed next fire in sequence during checkpoint processing.) arg0 is the number of buffers written. arg1 is the total number of buffers. arg2, arg3 and arg4 contain the number of WAL files added, removed and recycled respectively.

clog-checkpoint-start

(bool)

Probe that fires when the CLOG portion of a checkpoint is started. arg0 is true for normal checkpoint, false for shutdown checkpoint.

clog-checkpoint-done

(bool)

Probe that fires when the CLOG portion of a checkpoint is complete. arg0 has the same meaning as for clog-checkpoint-start.

subtrans-checkpoint-start

(bool)

Probe that fires when the SUBTRANS portion of a checkpoint is started. arg0 is true for normal checkpoint, false for shutdown checkpoint.

subtrans-checkpoint-done

(bool)

Probe that fires when the SUBTRANS portion of a checkpoint is complete. arg0 has the same meaning as for subtrans-checkpoint-start.

multixact-checkpoint-start

(bool)

Probe that fires when the MultiXact portion of a checkpoint is started. arg0 is true for normal checkpoint, false for shutdown checkpoint.

multixact-checkpoint-done

(bool)

Probe that fires when the MultiXact portion of a checkpoint is complete. arg0 has the same meaning as for multixact-checkpoint-start.

buffer-checkpoint-start

(int)

Probe that fires when the buffer-writing portion of a checkpoint is started. arg0 holds the bitwise flags used to distinguish different checkpoint types, such as shutdown, immediate or force.

buffer-sync-start

(int, int)

Probe that fires when we begin to write dirty buffers during checkpoint (after identifying which buffers must be written). arg0 is the total number of buffers. arg1 is the number that are currently dirty and need to be written.

buffer-sync-written

(int)

Probe that fires after each buffer is written during checkpoint. arg0 is the ID number of the buffer.

buffer-sync-done

(int, int, int)

Probe that fires when all dirty buffers have been written. arg0 is the total number of buffers. arg1 is the number of buffers actually written by the checkpoint process. arg2 is the number that were expected to be written (arg1 of buffer-sync-start); any difference reflects other processes flushing buffers during the checkpoint.

buffer-checkpoint-sync-start

()

Probe that fires after dirty buffers have been written to the kernel, and before starting to issue fsync requests.

buffer-checkpoint-done

()

Probe that fires when syncing of buffers to disk is complete.

twophase-checkpoint-start

()

Probe that fires when the two-phase portion of a checkpoint is started.

twophase-checkpoint-done

()

Probe that fires when the two-phase portion of a checkpoint is complete.

buffer-read-start

(ForkNumber, BlockNumber, Oid, Oid, Oid, int, bool)

Probe that fires when a buffer read is started. arg0 and arg1 contain the fork and block numbers of the page (but arg1 will be -1 if this is a relation extension request). arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation. arg5 is the ID of the backend which created the temporary relation for a local buffer, or InvalidBackendId (-1) for a shared buffer. arg6 is true for a relation extension request, false for normal read.

buffer-read-done

(ForkNumber, BlockNumber, Oid, Oid, Oid, int, bool, bool)

Probe that fires when a buffer read is complete. arg0 and arg1 contain the fork and block numbers of the page (if this is a relation extension request, arg1 now contains the block number of the newly added block). arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation. arg5 is the ID of the backend which created the temporary relation for a local buffer, or InvalidBackendId (-1) for a shared buffer. arg6 is true for a relation extension request, false for normal read. arg7 is true if the buffer was found in the pool, false if not.

buffer-flush-start

(ForkNumber, BlockNumber, Oid, Oid, Oid)

Probe that fires before issuing any write request for a shared buffer. arg0 and arg1 contain the fork and block numbers of the page. arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation.

buffer-flush-done

(ForkNumber, BlockNumber, Oid, Oid, Oid)

Probe that fires when a write request is complete. (Note that this just reflects the time to pass the data to the kernel; it's typically not actually been written to disk yet.) The arguments are the same as for buffer-flush-start.

buffer-write-dirty-start

(ForkNumber, BlockNumber, Oid, Oid, Oid)

Probe that fires when a server process begins to write a dirty buffer. (If this happens often, it implies that shared_buffers is too small or the background writer control parameters need adjustment.) arg0 and arg1 contain the fork and block numbers of the page. arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation.

buffer-write-dirty-done

(ForkNumber, BlockNumber, Oid, Oid, Oid)

Probe that fires when a dirty-buffer write is complete. The arguments are the same as for buffer-write-dirty-start.

wal-buffer-write-dirty-start

()

Probe that fires when a server process begins to write a dirty WAL buffer because no more WAL buffer space is available. (If this happens often, it implies that wal_buffers is too small.)

wal-buffer-write-dirty-done

()

Probe that fires when a dirty WAL buffer write is complete.

wal-insert

(unsigned char, unsigned char)

Probe that fires when a WAL record is inserted. arg0 is the resource manager (rmid) for the record. arg1 contains the info flags.

wal-switch

()

Probe that fires when a WAL segment switch is requested.

smgr-md-read-start

(ForkNumber, BlockNumber, Oid, Oid, Oid, int)

Probe that fires when beginning to read a block from a relation. arg0 and arg1 contain the fork and block numbers of the page. arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation. arg5 is the ID of the backend which created the temporary relation for a local buffer, or InvalidBackendId (-1) for a shared buffer.

smgr-md-read-done

(ForkNumber, BlockNumber, Oid, Oid, Oid, int, int, int)

Probe that fires when a block read is complete. arg0 and arg1 contain the fork and block numbers of the page. arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation. arg5 is the ID of the backend which created the temporary relation for a local buffer, or InvalidBackendId (-1) for a shared buffer. arg6 is the number of bytes actually read, while arg7 is the number requested (if these are different it indicates trouble).

smgr-md-write-start

(ForkNumber, BlockNumber, Oid, Oid, Oid, int)

Probe that fires when beginning to write a block to a relation. arg0 and arg1 contain the fork and block numbers of the page. arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation. arg5 is the ID of the backend which created the temporary relation for a local buffer, or InvalidBackendId (-1) for a shared buffer.

smgr-md-write-done

(ForkNumber, BlockNumber, Oid, Oid, Oid, int, int, int)

Probe that fires when a block write is complete. arg0 and arg1 contain the fork and block numbers of the page. arg2, arg3, and arg4 contain the tablespace, database, and relation OIDs identifying the relation. arg5 is the ID of the backend which created the temporary relation for a local buffer, or InvalidBackendId (-1) for a shared buffer. arg6 is the number of bytes actually written, while arg7 is the number requested (if these are different it indicates trouble).

sort-start

(int, bool, int, int, bool, int)

Probe that fires when a sort operation is started. arg0 indicates heap, index or datum sort. arg1 is true for unique-value enforcement. arg2 is the number of key columns. arg3 is the number of kilobytes of work memory allowed. arg4 is true if random access to the sort result is required. arg5 indicates serial when 0, parallel worker when 1, or parallel leader when 2.

sort-done

(bool, long)

Probe that fires when a sort is complete. arg0 is true for external sort, false for internal sort. arg1 is the number of disk blocks used for an external sort, or kilobytes of memory used for an internal sort.

lwlock-acquire

(char *, LWLockMode)

Probe that fires when an LWLock has been acquired. arg0 is the LWLock's tranche. arg1 is the requested lock mode, either exclusive or shared.

lwlock-release

(char *)

Probe that fires when an LWLock has been released (but note that any released waiters have not yet been awakened). arg0 is the LWLock's tranche.

lwlock-wait-start

(char *, LWLockMode)

Probe that fires when an LWLock was not immediately available and a server process has begun to wait for the lock to become available. arg0 is the LWLock's tranche. arg1 is the requested lock mode, either exclusive or shared.

lwlock-wait-done

(char *, LWLockMode)

Probe that fires when a server process has been released from its wait for an LWLock (it does not actually have the lock yet). arg0 is the LWLock's tranche. arg1 is the requested lock mode, either exclusive or shared.

lwlock-condacquire

(char *, LWLockMode)

Probe that fires when an LWLock was successfully acquired when the caller specified no waiting. arg0 is the LWLock's tranche. arg1 is the requested lock mode, either exclusive or shared.

lwlock-condacquire-fail

(char *, LWLockMode)

Probe that fires when an LWLock was not successfully acquired when the caller specified no waiting. arg0 is the LWLock's tranche. arg1 is the requested lock mode, either exclusive or shared.

lock-wait-start

(unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, LOCKMODE)

Probe that fires when a request for a heavyweight lock (lmgr lock) has begun to wait because the lock is not available. arg0 through arg3 are the tag fields identifying the object being locked. arg4 indicates the type of object being locked. arg5 indicates the lock type being requested.

lock-wait-done

(unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, LOCKMODE)

Probe that fires when a request for a heavyweight lock (lmgr lock) has finished waiting (i.e., has acquired the lock). The arguments are the same as for lock-wait-start.

deadlock-found

()

Probe that fires when a deadlock is found by the deadlock detector.

Table 28.48. Defined Types Used in Probe Parameters

TypeDefinition

LocalTransactionId

unsigned int

LWLockMode

int

LOCKMODE

int

BlockNumber

unsigned int

Oid

unsigned int

ForkNumber

int

bool

unsigned char

28.5.3. Using Probes

The example below shows a DTrace script for analyzing transaction counts in the system, as an alternative to snapshotting pg_stat_database before and after a performance test:

#!/usr/sbin/dtrace -qs

postgresql$1:::transaction-start
{
      @start["Start"] = count();
      self->ts  = timestamp;
}

postgresql$1:::transaction-abort
{
      @abort["Abort"] = count();
}

postgresql$1:::transaction-commit
/self->ts/
{
      @commit["Commit"] = count();
      @time["Total time (ns)"] = sum(timestamp - self->ts);
      self->ts=0;
}

When executed, the example D script gives output such as:

# ./txn_count.d `pgrep -n postgres` or ./txn_count.d <PID>
^C

Start                                          71
Commit                                         70
Total time (ns)                        2312105013

Note

SystemTap uses a different notation for trace scripts than DTrace does, even though the underlying trace points are compatible. One point worth noting is that at this writing, SystemTap scripts must reference probe names using double underscores in place of hyphens. This is expected to be fixed in future SystemTap releases.

You should remember that DTrace scripts need to be carefully written and debugged, otherwise the trace information collected might be meaningless. In most cases where problems are found it is the instrumentation that is at fault, not the underlying system. When discussing information found using dynamic tracing, be sure to enclose the script used to allow that too to be checked and discussed.

28.5.4. Defining New Probes

New probes can be defined within the code wherever the developer desires, though this will require a recompilation. Below are the steps for inserting new probes:

  1. Decide on probe names and data to be made available through the probes

  2. Add the probe definitions to src/backend/utils/probes.d

  3. Include pg_trace.h if it is not already present in the module(s) containing the probe points, and insert TRACE_POSTGRESQL probe macros at the desired locations in the source code

  4. Recompile and verify that the new probes are available

Example: Here is an example of how you would add a probe to trace all new transactions by transaction ID.

  1. Decide that the probe will be named transaction-start and requires a parameter of type LocalTransactionId

  2. Add the probe definition to src/backend/utils/probes.d:

    probe transaction__start(LocalTransactionId);

    Note the use of the double underline in the probe name. In a DTrace script using the probe, the double underline needs to be replaced with a hyphen, so transaction-start is the name to document for users.

  3. At compile time, transaction__start is converted to a macro called TRACE_POSTGRESQL_TRANSACTION_START (notice the underscores are single here), which is available by including pg_trace.h. Add the macro call to the appropriate location in the source code. In this case, it looks like the following:

    TRACE_POSTGRESQL_TRANSACTION_START(vxid.localTransactionId);
  4. After recompiling and running the new binary, check that your newly added probe is available by executing the following DTrace command. You should see similar output:

    # dtrace -ln transaction-start
       ID    PROVIDER          MODULE           FUNCTION NAME
    18705 postgresql49878     postgres     StartTransactionCommand transaction-start
    18755 postgresql49877     postgres     StartTransactionCommand transaction-start
    18805 postgresql49876     postgres     StartTransactionCommand transaction-start
    18855 postgresql49875     postgres     StartTransactionCommand transaction-start
    18986 postgresql49873     postgres     StartTransactionCommand transaction-start

There are a few things to be careful about when adding trace macros to the C code:

  • You should take care that the data types specified for a probe's parameters match the data types of the variables used in the macro. Otherwise, you will get compilation errors.

  • On most platforms, if PostgreSQL is built with --enable-dtrace, the arguments to a trace macro will be evaluated whenever control passes through the macro, even if no tracing is being done. This is usually not worth worrying about if you are just reporting the values of a few local variables. But beware of putting expensive function calls into the arguments. If you need to do that, consider protecting the macro with a check to see if the trace is actually enabled:

    if (TRACE_POSTGRESQL_TRANSACTION_START_ENABLED())
        TRACE_POSTGRESQL_TRANSACTION_START(some_function(...));

    Each trace macro has a corresponding ENABLED macro.