使用 gdb 调试 postgres greenplum
错误信息的获取途径有几种 :
1. 最简单的就是看PostgreSQL的log文件.
2. 在PostgreSQL中可以使用psql的 \set VERBOSITY verbose 来输出详细的错误信息,
3. 或者使用linux的strace跟踪这个发生错误的进程.
4. 使用gdb (GNU Debugger) 来调试则是一种比较深入但是非常有用的工具.
下面就以前面我在使用pgsql_fdw时遇到的问题使用gdb来做个简单的演示 :
要让postgresql能够使用gdb进行调试, 首先要在编译PostgreSQL时打开enable-debug.
例如, 使用如下方法编译 :
. /configure --prefix=/ home / digoal / pgsql91 -- with - pgport = 1999 -- with - perl -- with - python -- with - openssl -- with - pam -- without - ldap -- with - libxml -- with - libxslt -- enable - thread - safety -- with - wal - blocksize = 16 -- enable - debug && gmake worldsudo gmake install - world
在满足这个条件后, 初始化数据库
initdb -D $PGDATA -E UTF8 --locale=C -W -U postgres
使用gdb进行调试 :
会话1 :
获取当前进程的pid.
digoal@db - 172 - 16 - 3 - 33 -> psql postgres postgrespsql ( 9.1 . 3 )Type "help" for help .
postgres = # select pg_backend_pid();pg_backend_pid----------------26841(1 row)
postgres = # \set VERBOSITY verbose
会话2 :
使用gdb, 绑定到会话1获取到的进程PID.
digoal@db - 172 - 16 - 3 - 33 -> gdb( gdb ) attach 26841
在出错的函数处设置断点(
b getTypeOutputInfo) :
( gdb ) b getTypeOutputInfoBreakpoint 1 at 0x6cbaf0 : file lsyscache . c , line 2434.
会话1 :
执行那个报错的函数 :
postgres=# select f_test();
会话2 :
继续执行(c), 跟踪到
getTypeOutputInfo的输入参数如下 :
( gdb ) cContinuing .Breakpoint 1 , getTypeOutputInfo ( type = 0 , typOutput = 0x7fff1c55df58 , typIsVarlena = 0x7fff1c55df5f "" ) at lsyscache . c : 24342434 {
查看截止到断点前的所有调用(bt或backtrace) :
注意这里的结果和实际的调用顺序是反的, 如, #1的execute_query调用了#0的
getTypeOutputInfo
( gdb ) bt#0 getTypeOutputInfo (type=0, typOutput=0x7fff1c55df58, typIsVarlena=0x7fff1c55df5f "") at lsyscache.c:2434#1 0x00002b066328943b in execute_query (node=0x17f14c70) at pgsql_fdw.c:668#2 0x00002b066328a00f in pgsqlIterateForeignScan (node=0x0) at pgsql_fdw.c:352#3 0x0000000000573a6f in ForeignNext (node=0x0) at nodeForeignscan.c:49#4 0x0000000000560b1e in ExecScanFetch (node=0x17f14c70, accessMtd=0x573a30 <ForeignNext>, recheckMtd=0x5738b0 <ForeignRecheck>)at execScan.c:82#5 ExecScan (node=0x17f14c70, accessMtd=0x573a30 <ForeignNext>, recheckMtd=0x5738b0 <ForeignRecheck>) at execScan.c:167#6 0x0000000000559dda in ExecProcNode (node=0x17f14c70) at execProcnode.c:432#7 0x000000000056b5c8 in ExecLimit (node=0x17f14970) at nodeLimit.c:91#8 0x0000000000559eaa in ExecProcNode (node=0x17f14970) at execProcnode.c:490#9 0x0000000000558d21 in ExecutePlan (queryDesc=0x17f12850, direction=475389784, count=0) at execMain.c:1439#10 standard_ExecutorRun (queryDesc=0x17f12850, direction=475389784, count=0) at execMain.c:313#11 0x00000000005788a3 in _SPI_execute_plan (plan=0x17f16870, paramLI=0x17e0ae68, snapshot=<value optimized out>,crosscheck_snapshot=0x0, read_only=0 '\000', fire_triggers=1 '\001', tcount=0) at spi.c:2110#12 0x0000000000578c9d in SPI_execute_plan_with_paramlist (plan=0x17f16870, params=0x17e0ae68, read_only=0 '\000', tcount=0)at spi.c:423#13 0x00002b0663072133 in exec_run_select (estate=0x7fff1c55e580, expr=0x17f0c880, maxtuples=0, portalP=0x0) at pl_exec.c:4580#14 0x00002b0663075809 in exec_stmt (estate=0x7fff1c55e580, stmts=<value optimized out>) at pl_exec.c:1413#15 exec_stmts (estate=0x7fff1c55e580, stmts=<value optimized out>) at pl_exec.c:1248#16 0x00002b0663075628 in exec_stmt_block (estate=0x7fff1c55e580, block=0x17f0cb40) at pl_exec.c:1186#17 0x00002b0663078273 in plpgsql_exec_function (func=0x17eee148, fcinfo=0x17f18940) at pl_exec.c:324#18 0x00002b066306d3e3 in plpgsql_call_handler (fcinfo=0x17f18940) at pl_handler.c:122#19 0x000000000055d5be in ExecMakeFunctionResult (fcache=0x17f188d0, econtext=0x17f186e0, isNull=0x17f19258 "\024",isDone=0x17f19370) at execQual.c:1824#20 0x000000000055a35a in ExecTargetList (projInfo=<value optimized out>, isDone=0x7fff1c55e92c) at execQual.c:5104#21 ExecProject (projInfo=<value optimized out>, isDone=0x7fff1c55e92c) at execQual.c:5319#22 0x000000000056fcd3 in ExecResult (node=0x17f185d0) at nodeResult.c:155#23 0x0000000000559d06 in ExecProcNode (node=0x17f185d0) at execProcnode.c:367#24 0x0000000000558d21 in ExecutePlan (queryDesc=0x17ebbfa0, direction=475389784, count=0) at execMain.c:1439#25 standard_ExecutorRun (queryDesc=0x17ebbfa0, direction=475389784, count=0) at execMain.c:313#26 0x000000000061d9d4 in PortalRunSelect (portal=0x17eb9f90, forward=<value optimized out>, count=0, dest=0x17eb0050)at pquery.c:943#27 0x000000000061eee0 in PortalRun (portal=0x17eb9f90, count=9223372036854775807, isTopLevel=1 '\001', dest=0x17eb0050,altdest=0x17eb0050, completionTag=0x7fff1c55eb90 "") at pquery.c:787#28 0x000000000061bbbb in exec_simple_query (query_string=0x17eae9c0 "select f_test();") at postgres.c:1018#29 0x000000000061c3c4 in PostgresMain (argc=<value optimized out>, argv=<value optimized out>, username=<value optimized out>)at postgres.c:3926#30 0x00000000005e191b in ServerLoop () at postmaster.c:3606#31 0x00000000005e253c in PostmasterMain (argc=1, argv=0x17e05090) at postmaster.c:1116---Type <return> to continue, or q <return> to quit---#32 0x0000000000586d6e in main (argc=1, argv=<value optimized out>) at main.c:199
其他常用gdb指令 :
print
next
nexti
【参考】
1. (gdb) 是gdb的命令行提示符.
2. http://en.wikipedia.org/wiki/GNU_Debugger
3. http://wiki.postgresql.org/wiki/Pgsrcstructure
4. gdb 所有指令 :
(gdb) help all