我的书签
添加书签
移除书签背景
数据库的性能优化是一个非常经典的话题，数据库的优化手段以及优化的角度也各不相同。
例如，可以从OS内核、网络、块设备、编译器、文件系统、SQL、数据库参数、业务逻辑、源码等各个方面去进行优化。
但是如果在优化前了解瓶颈在什么地方，可以向打鼹鼠一样，先打大的，起到事半功倍的效果。
本文将针对Linux平台下的PostgreSQL服务，讲解如何诊断PostgreSQL 数据库服务的瓶颈，了解瓶颈在哪个方面，甚至可以了解到瓶颈在哪段代码？
有了比较详细的性能诊断报告，可以给出比较好的指导优化建议。
我们要用到的工具是 《Linux 性能诊断(profiling) perf使用指南》
使用perf跟踪PostgreSQL的前提
要输出完备的跟踪信息，如符号表，call stack traces, 汇编指令。必须在编译PostgreSQL时设置相应的编译开关，例如GCC需要开启如下
1. CFLAGS="-g -ggdb -fno-omit-frame-pointer"
建议Linux内核编译时加上
1. CONFIG_KALLSYMS=y
2. CONFIG_FRAME_POINTER=y
编译perf时需要支持libunwind, 并加上
1. gcc -g dwarf
3. man gcc
4. -gdwarf-version
5. Produce debugging information in DWARF format (if that is supported). This is the format used by DBX on IRIX 6. The value of version may be either 2 or 3; the default version is 3.
7. Note that with DWARF version 2 some ports require, and will always use, some non-conflicting DWARF 3 extensions in the unwind tables.
如果是yum安装的软件，可以安装对应的debuginfo包。
例子
1. wget https://ftp.postgresql.org/pub/source/v9.6.1/postgresql-9.6.1.tar.bz2
2. tar -jxvf postgresql-9.6.1.tar.bz2
4. cd postgresql-9.6.1
5. export USE_NAMED_POSIX_SEMAPHORES=1
6. LIBS=-lpthread CC="/home/digoal/gcc6.2.0/bin/gcc" CFLAGS="-O3 -flto -g -ggdb -fno-omit-frame-pointer" ./configure --prefix=/home/digoal/pgsql9.6
7. LIBS=-lpthread CC="/home/digoal/gcc6.2.0/bin/gcc" CFLAGS="-O3 -flto -g -ggdb -fno-omit-frame-pointer" make world -j 64
8. LIBS=-lpthread CC="/home/digoal/gcc6.2.0/bin/gcc" CFLAGS="-O3 -flto -g -ggdb -fno-omit-frame-pointer" make install-world
常用跟踪手段
1. benchmark
首先，需要定一个目标，比如insert only的场景，如何找到其瓶颈。
创建测试表
1. postgres=# create table perf_test_ins(id serial8 primary key, info text, crt_time timestamp);
2. alCREATE TABLE
3. postgres=# alter sequence perf_test_ins_id_seq cache 10000;
4. ALTER SEQUENCE
创建测试脚本，不停插入
1. vi test.sql
3. insert into perf_test_ins(info,crt_time) values ('test',now());
疲劳压测开始
1. pgbench -M prepared -n -r -P 1 -f ./test.sql -h $PGDATA -c 32 -j 32 -T 600
2. 实时跟踪
pgbench压测的同时，实时观测。
1. su - root
3. perf top -agv -F 1000
6. 解释
7. perf top --help
9. NAME
10. perf-top - System profiling tool.
12. SYNOPSIS
13. perf top [-e <EVENT> | --event=EVENT] [<options>]
15. DESCRIPTION
16. This command generates and displays a performance counter profile in real time.
18. -F <freq>, --freq=<freq>
19. Profile at this frequency.
20. -a, --all-cpus
21. System-wide collection. (default)
22. -g, --group
23. Put the counters into a counter group.
24. -K, --hide_kernel_symbols
25. Hide kernel symbols.
26. -U, --hide_user_symbols
27. Hide user symbols.
28. -v, --verbose
29. Be more verbose (show counter open errors, etc).
输出
1. PerfTop: 23287 irqs/sec kernel:31.2% exact: 0.0% [1000Hz cycles], (all, 24 CPUs)
2. --------------------------------------------------------------------------------------------------------------
4. samples pcnt RIP function DSO
5. _______ _____ ________________ _________________________________________ _____________________________________
7. 4179.00 1.6% 0000000000524e10 AllocSetAlloc.lto_priv.1186 /home/digoal/pgsql9.6/bin/postgres
8. 3966.00 1.5% 0000000000043d10 __GI_vfprintf /lib64/libc-2.12.so
9. 3822.00 1.5% ffffffff8150bb50 schedule [kernel.kallsyms]
10. 3659.00 1.4% 0000000000082e10 __memset_sse2 /lib64/libc-2.12.so
11. 3523.00 1.4% 00000000005782d0 LWLockAcquire.constprop.931 /home/digoal/pgsql9.6/bin/postgres
12. 3219.00 1.2% 00000000002c9390 LWLockRelease /home/digoal/pgsql9.6/bin/postgres
13. 3114.00 1.2% 0000000000332e10 PostgresMain /home/digoal/pgsql9.6/bin/postgres
14. 3042.00 1.2% ffffffff8150e8b0 _spin_lock [kernel.kallsyms]
15. 2948.00 1.1% 0000000000573fe0 hash_search_with_hash_value.constprop.930 /home/digoal/pgsql9.6/bin/postgres
16. 2901.00 1.1% 00000000005746d0 hash_search_with_hash_value.constprop.928 /home/digoal/pgsql9.6/bin/postgres
17. 2848.00 1.1% 00000000002db5f0 GetSnapshotData /home/digoal/pgsql9.6/bin/postgres
18. 2471.00 1.0% 00000000005338b0 hash_seq_search /home/digoal/pgsql9.6/bin/postgres
19. 2275.00 0.9% 0000000000579ad0 LWLockAcquire.constprop.932 /home/digoal/pgsql9.6/bin/postgres
20. 2244.00 0.9% 0000000000078460 _int_malloc /lib64/libc-2.12.so
21. 2215.00 0.9% 0000000000088aa0 memcpy /lib64/libc-2.12.so
22. 2163.00 0.8% 00000000002df050 LockAcquireExtended /home/digoal/pgsql9.6/bin/postgres
23. 2057.00 0.8% 0000000000131a70 __strlen_sse42 /lib64/libc-2.12.so
24. 2017.00 0.8% 00000000002ba2d0 _bt_compare /home/digoal/pgsql9.6/bin/postgres
25. 1977.00 0.8% 0000000000006e40 doCustom.lto_priv.13 /home/digoal/pgsql9.6/bin/pgbench
26. 1951.00 0.8% ffffffff8100a630 __switch_to [kernel.kallsyms]
27. 1855.00 0.7% 000000000053d5e0 hash_search_with_hash_value /home/digoal/pgsql9.6/bin/postgres
28. 1767.00 0.7% 00000000001c6670 hash_any /home/digoal/pgsql9.6/bin/postgres
29. 1726.00 0.7% 00000000002c8310 PinBuffer.isra.3.lto_priv.2340 /home/digoal/pgsql9.6/bin/postgres
30. 1668.00 0.6% 00000000000ee570 XLogInsertRecord /home/digoal/pgsql9.6/bin/postgres
31. 1604.00 0.6% 0000000000250500 pgstat_report_activity /home/digoal/pgsql9.6/bin/postgres
32. 1572.00 0.6% ffffffff81189c00 fget_light [kernel.kallsyms]
33. 1565.00 0.6% 0000000000426970 SearchCatCache /home/digoal/pgsql9.6/bin/postgres
34. 1460.00 0.6% ffffffff810e03a0 __audit_syscall_exit [kernel.kallsyms]
35. 1415.00 0.5% ffffffff81057e80 update_curr [kernel.kallsyms]
36. 1409.00 0.5% 00000000000777f0 _int_free /lib64/libc-2.12.so
37. 1386.00 0.5% 00000000000fbc10 CommitTransaction.lto_priv.2946 /home/digoal/pgsql9.6/bin/postgres
38. 1296.00 0.5% ffffffff8119f6d0 do_select [kernel.kallsyms]
39. 1285.00 0.5% 00000000002c8690 LWLockWaitListLock.lto_priv.1265 /home/digoal/pgsql9.6/bin/postgres
40. 1268.00 0.5% ffffffff8150e790 _spin_lock_irqsave [kernel.kallsyms]
41. 1247.00 0.5% 0000000000577650 hash_search_with_hash_value.constprop.929 /home/digoal/pgsql9.6/bin/postgres
42. 1220.00 0.5% 0000000000538a10 AllocSetFree.lto_priv.1187 /home/digoal/pgsql9.6/bin/postgres
43. 1201.00 0.5% 0000000000389660 ExecModifyTable /home/digoal/pgsql9.6/bin/postgres
44. 1185.00 0.5% 000000000009b6d0 gettimeofday /lib64/libc-2.12.so
45. 1171.00 0.5% ffffffff81278a60 copy_user_generic_string [kernel.kallsyms]
46. 1142.00 0.4% 000000000013cb10 __memcmp_sse4_1 /lib64/libc-2.12.so
47. 1124.00 0.4% 0000000000032620 __GI___sigsetjmp /lib64/libc-2.12.so
48. 1123.00 0.4% 00000000000ef2b0 XLogInsert /home/digoal/pgsql9.6/bin/postgres
49. 1112.00 0.4% 00000000002d29e0 LockReleaseAll /home/digoal/pgsql9.6/bin/postgres
50. 1108.00 0.4% 00000000003c7620 ExecInitExpr /home/digoal/pgsql9.6/bin/postgres
51. 1056.00 0.4% 000000000000ac80 threadRun.lto_priv.20 /home/digoal/pgsql9.6/bin/pgbench
52. 1052.00 0.4% ffffffff81062a40 select_task_rq_fair [kernel.kallsyms]
53. 1033.00 0.4% 0000000000017710 pqParseInput3 /home/digoal/pgsql9.6/lib/libpq.so.5.9
按Enter键弹出可以配置的选项以及当前状态。
1. Mapped keys:
2. [d] display refresh delay. (2)
3. [e] display entries (lines). (47)
4. [f] profile display filter (count). (5)
5. [F] annotate display filter (percent). (5%)
6. [s] annotate symbol. (NULL)
7. [S] stop annotation.
8. [K] hide kernel_symbols symbols. (no) 是否显示kernel symbols
9. [U] hide user symbols. (no) 是否显示user symbols
10. [z] toggle sample zeroing. (0)
11. [qQ] quit.
调整压测命令，连接回环地址，(前面连接的是unix socket)
1. pgbench -M prepared -n -r -P 1 -f ./test.sql -h 127.0.0.1 -c 32 -j 32 -T 600
再次观测perf top结果, 采样消耗排名第一的变成了这条，找到了一个瓶颈，关闭iptables，这条就消失了，TPS也提升了。
1. 5871.00 1.7% 0000000000002e40 ipt_do_table /lib/modules/2.6.32-358.23.2.ali1195.el6.x86_64/kernel/net/ipv4/netfilter/ip_tables.ko
3. 跟踪事件计数（指定事件）
指定命令进行跟踪，或者指定PID进行跟踪。
1. NAME
2. perf-stat - Run a command and gather performance counter statistics
4. SYNOPSIS
5. perf stat [-e <EVENT> | --event=EVENT] [-a] <command>
6. perf stat [-e <EVENT> | --event=EVENT] [-a] — <command> [<options>]
8. DESCRIPTION
9. This command runs a command and gathers performance counter statistics from it.
10. -a, --all-cpus
11. system-wide collection from all CPUs
12. -c, --scale
13. scale/normalize counter values
14. -B, --big-num
15. print large numbers with thousands′ separators according to locale
16. -v, --verbose
17. be more verbose (show counter open errors, etc)
18. -p, --pid=<pid>
19. stat events on existing process id
20. -t, --tid=<tid>
21. stat events on existing thread id
22. -r, --repeat=<n>
23. repeat command and print average + stddev (max: 100)
跟踪批量插入一批记录的事件调用统计
1. event来自perf list的输出
3. perf stat -e block:*,syscalls:*,xfs:* -a -B -v -c -- /home/digoal/pgsql9.6/bin/psql -h 127.0.0.1 -p 5288 -q -U postgres postgres -c "drop table if exists x; create table x as select a FROM generate_series(1,1000000) a;";
输出
1. 0 block:block_rq_abort [98.79%]
2. 0 block:block_rq_requeue [98.79%]
3. 995 block:block_rq_complete [98.79%]
4. 993 block:block_rq_insert [98.79%]
5. 997 block:block_rq_issue [98.79%]
6. 0 block:block_bio_bounce [98.79%]
7. 0 block:block_bio_complete [98.79%]
8. 18 block:block_bio_backmerge [98.79%]
9. 1 block:block_bio_frontmerge [98.79%]
10. 1,011 block:block_bio_queue [98.79%]
11. 992 block:block_getrq [98.79%]
12. 2 block:block_sleeprq [98.79%]
13. 250 block:block_plug [98.79%]
14. 2 block:block_unplug_timer [98.79%]
15. 361 block:block_unplug_io [98.79%]
16. 0 block:block_split [98.79%]
17. 1,011 block:block_remap [98.79%]
18. 0 block:block_rq_remap [98.79%]
19. 5 syscalls:sys_enter_socket [98.79%]
20. 5 syscalls:sys_exit_socket [98.79%]
21. 0 syscalls:sys_enter_socketpair [98.79%]
22. 0 syscalls:sys_exit_socketpair [98.79%]
23. ... 略 ...
计数例子
1. perf stat -a -B -v -c -- /home/digoal/pgsql9.6/bin/psql -h 127.0.0.1 -p 5288 -q -U postgres postgres -c "drop table if exists x; create table x as select a FROM generate_series(1,1000000) a;";
3. 22433.920027 task-clock # 24.042 CPUs utilized [100.00%]
4. 4,275 context-switches # 0.000 M/sec [100.00%]
5. 98 CPU-migrations # 0.000 M/sec [100.00%]
6. 56,484 page-faults # 0.003 M/sec
7. 2,687,623,473 cycles # 0.120 GHz [80.73%]
8. 2,747,044,652 stalled-cycles-frontend # 102.21% frontend cycles idle [92.42%]
9. 2,358,755,489 stalled-cycles-backend # 87.76% backend cycles idle [81.89%]
10. 3,373,693,967 instructions # 1.26 insns per cycle 越高越好，说明CPU没闲着。至少要大于1，但是注意同步锁等待这个值也可能很高，CPU空转，这就不好了。
11. # 0.81 stalled cycles per insn [91.50%] 越低越好，说明CPU没有闲着。
12. 632,351,270 branches # 28.187 M/sec [91.82%]
13. 1,236,798 branch-misses # 0.20% of all branches [53.19%]
15. 0.933128299 seconds time elapsed
perf stat重点看的是instructions，判断CPU是否充分被利用，如果没有，说明系统有其他瓶颈，例如IO等待，或者其他等待。
4. 采样跟踪
这个是用得最多的，先采样，后分析报告。
1. NAME
2. perf-record - Run a command and record its profile into perf.data
4. SYNOPSIS
5. perf record [-e <EVENT> | --event=EVENT] [-l] [-a] <command>
6. perf record [-e <EVENT> | --event=EVENT] [-l] [-a] — <command> [<options>]
8. DESCRIPTION
9. This command runs a command and gathers a performance counter profile from it, into perf.data - without displaying anything.
11. This file can then be inspected later on, using perf report.
13. -A, --append
14. Append to the output file to do incremental profiling.
15. -o, --output=
16. Output file name.
17. -e, --event= event来自perf list的输出
18. Select the PMU event.
19. Selection can be a symbolic event name (use perf list to list all events) or a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a hexadecimal event descriptor.
20. -a, --all-cpus
21. System-wide collection from all CPUs.
22. -g, --call-graph
23. Do call-graph (stack chain/backtrace) recording.
24. -v, --verbose
25. Be more verbose (show counter open errors, etc).
26. -s, --stat 相对的是默认的per cpu mode
27. Per thread counts.
28. -G name,..., --cgroup name,... 如果你只想跟踪某一些进程的统计信息，可以将这些进程放到一个cgroup中，跟踪这个cgroup的。
29. monitor only in the container (cgroup) called "name".
30. This option is available only in per-cpu mode.
31. The cgroup filesystem must be mounted.
32. All threads belonging to container "name" are monitored when they run on the monitored CPUs.
33. Multiple cgroups can be provided.
34. Each cgroup is applied to the corresponding event, i.e., first cgroup to first event, second cgroup to second event and so on. It is
35. possible to provide an empty cgroup (monitor all the time) using, e.g., -G foo,,bar.
36. Cgroups must have corresponding events, i.e., they always refer to events defined earlier on the command line.
跑pgbench压测，和前面的pgbench一样，不断并发插入。
然后开启统计收集，收集10秒的统计信息，默认输出到perf.data文件。
1. perf record -avg -- sleep 10
3. [ perf record: Woken up 142 times to write data ]
4. [ perf record: Captured and wrote 37.111 MB perf.data (~1621415 samples) ]
5. 生成报告
前面使用perf record收集了统计信息到perf.data，接下来就对perf.data进行分析，输出报告。
1. NAME
2. perf-report - Read perf.data (created by perf record) and display the profile
4. SYNOPSIS
5. perf report [-i <file> | --input=file]
7. DESCRIPTION
8. This command displays the performance counter profile information recorded via perf record.
10. -i, --input=
11. Input file name. (default: perf.data)
12. -v, --verbose
13. Be more verbose. (show symbol address, etc)
14. -n, --show-nr-samples
15. Show the number of samples for each symbol
16. --showcpuutilization
17. Show sample percentage for different cpu modes.
18. -T, --threads
19. Show per-thread event counters
20. -U, --hide-unresolved
21. Only display entries resolved to a symbol.
22. -s, --sort=
23. Sort by key(s): pid, comm, dso, symbol, parent.
24. -g [type,min], --call-graph
25. Display call chains using type and min percent threshold. type can be either:
26. · flat: single column, linear exposure of call chains.
27. · graph: use a graph tree, displaying absolute overhead rates.
28. · fractal: like graph, but displays relative rates. Each branch of the tree is considered as a new profiled object.
29. Default: fractal,0.5.
30. --stdio
31. Use the stdio interface.
32. --tui
33. Use the TUI interface, that is integrated with annotate and allows zooming into DSOs or threads, among other features. Use of --tui requires a tty, if one is not present, as when piping to other
34. commands, the stdio interface is used.
--tui模式可以查看概览，支持交互式指令。
1. 输入E全展开，输入C全收敛。 展开后可以使用键盘的上下键进行浏览。
2. [.] 表示userspace
3. [k] 表示kernel
5. 1.39% 0.00% 0.01% 15 postgres /home/digoal/pgsql9.6/bin/postgres 0x524e3a d [.] AllocSetAlloc.lto_priv.1186
6. 1.29% 0.00% 0.01% 22 postgres /lib64/libc-2.12.so 0x83127 B [.] __memset_sse2
7. 1.25% 0.00% 0.00% 1 postgres /home/digoal/pgsql9.6/bin/postgres 0x57831d d [.] LWLockAcquire.constprop.931
8. 1.24% 0.00% 0.05% 107 postgres /home/digoal/pgsql9.6/bin/postgres 0x574751 d [.] hash_search_with_hash_value.constprop.928
9. 1.07% 0.00% 0.00% 2 postgres /home/digoal/pgsql9.6/bin/postgres 0x2c946f d [.] LWLockRelease
10. 1.01% 0.00% 0.00% 1 postgres /home/digoal/pgsql9.6/bin/postgres 0x574035 d [.] hash_search_with_hash_value.constprop.930
11. 0.98% 0.00% 0.03% 74 postgres /home/digoal/pgsql9.6/bin/postgres 0x334832 d [.] PostgresMain
12. 0.94% 0.00% 0.02% 49 postgres /home/digoal/pgsql9.6/bin/postgres 0x2db87b d [.] GetSnapshotData
13. 0.94% 0.00% 0.00% 1 postgres [kernel.kallsyms] 0xffffffff8150e8b1 k [k] _spin_lock
14. 0.83% 0.00% 0.00% 1 postgres [kernel.kallsyms] 0xffffffff8150edb0 k [k] page_fault
15. 0.81% 0.00% 0.00% 1 postgres /lib64/libc-2.12.so 0x88aa0 B [.] memcpy
16. 0.77% 0.00% 0.22% 465 postgres /home/digoal/pgsql9.6/bin/postgres 0x533939 d [.] hash_seq_search
17. 0.75% 0.00% 0.00% 1 postgres /lib64/libc-2.12.so 0x454ac B [.] __GI_vfprintf
18. 0.75% 0.00% 0.03% 59 postgres /home/digoal/pgsql9.6/bin/postgres 0x579b1d d [.] LWLockAcquire.constprop.932
19. 0.73% 0.00% 0.02% 42 postgres /home/digoal/pgsql9.6/bin/postgres 0x2df3ab d [.] LockAcquireExtended
20. 0.67% 0.00% 0.00% 2 postgres /home/digoal/pgsql9.6/bin/postgres 0x1c66f9 d [.] hash_any
21. 0.65% 0.00% 0.02% 49 postgres /home/digoal/pgsql9.6/bin/postgres 0x2ba33a d [.] _bt_compare
22. 0.63% 0.00% 0.00% 3 postgres /home/digoal/pgsql9.6/bin/postgres 0xee632 d [.] XLogInsertRecord
23. 0.62% 0.00% 0.00% 1 postgres /home/digoal/pgsql9.6/bin/postgres 0x53d5e1 d [.] hash_search_with_hash_value
24. 0.59% 0.00% 0.59% 1318 pgbench /lib64/libc-2.12.so 0x47956 B [.] __GI_vfprintf
25. 0.57% 0.00% 0.00% 1 postgres /lib64/libc-2.12.so 0x131aa7 B [.] __strlen_sse42
26. 0.55% 0.00% 0.55% 1241 pgbench /home/digoal/pgsql9.6/bin/pgbench 0x85b3 d [.] doCustom.lto_priv.13
27. 0.55% 0.00% 0.03% 55 postgres /home/digoal/pgsql9.6/bin/postgres 0x2c8327 d [.] PinBuffer.isra.3.lto_priv.2340
28. 0.54% 0.00% 0.02% 35 postgres /home/digoal/pgsql9.6/bin/postgres 0x25059f d [.] pgstat_report_activity
29. 0.53% 0.00% 0.02% 43 postgres /lib64/libc-2.12.so 0x78462 B [.] _int_malloc
30. 0.52% 0.00% 0.01% 33 postgres /home/digoal/pgsql9.6/bin/postgres 0x426a36 d [.] SearchCatCache
31. 0.47% 0.00% 0.00% 3 postgres [kernel.kallsyms] 0xffffffff81272a1c k [k] radix_tree_lookup_slot
32. 0.45% 0.00% 0.01% 27 postgres /home/digoal/pgsql9.6/bin/postgres 0xfbe51 d [.] CommitTransaction.lto_priv.2946
33. 0.44% 0.00% 0.00% 1 postgres /home/digoal/pgsql9.6/bin/postgres 0x5776f5 d [.] hash_search_with_hash_value.constprop.929
34. 0.43% 0.00% 0.01% 30 postgres /home/digoal/pgsql9.6/bin/postgres 0x38a044 d [.] ExecModifyTable
35. 0.41% 0.00% 0.00% 1 postgres [kernel.kallsyms] 0xffffffff8150bef4 k [k] schedule
36. 0.40% 0.00% 0.01% 22 postgres /home/digoal/pgsql9.6/bin/postgres 0x2c86c2 d [.] LWLockWaitListLock.lto_priv.1265
37. 0.40% 0.00% 0.01% 29 postgres /home/digoal/pgsql9.6/bin/postgres 0x2d2b2d d [.] LockReleaseAll
38. 0.40% 0.00% 0.01% 12 postgres /home/digoal/pgsql9.6/bin/postgres 0x538a11 d [.] AllocSetFree.lto_priv.1187
39. 0.39% 0.00% 0.00% 1 postgres [kernel.kallsyms] 0xffffffff8111e567 k [k] find_get_page
40. 0.39% 0.01% 0.00% 24 postgres [kernel.kallsyms] 0xffffffff8148d1f8 k [k] tcp_sendmsg
41. 0.39% 0.00% 0.01% 34 postgres /home/digoal/pgsql9.6/bin/postgres 0xef545 d [.] XLogInsert
42. 0.39% 0.00% 0.01% 25 postgres /lib64/libc-2.12.so 0x32623 B [.] __GI___sigsetjmp
43. 0.38% 0.00% 0.38% 860 pgbench /home/digoal/pgsql9.6/bin/pgbench 0xb3f3 d [.] threadRun.lto_priv.20
44. 0.38% 0.00% 0.01% 33 postgres /lib64/libc-2.12.so 0x77cf2 B [.] _int_free
45. 0.38% 0.00% 0.00% 1 postgres /lib64/libc-2.12.so 0x13cb10 B [.] __memcmp_sse4_1
46. 0.38% 0.00% 0.02% 36 postgres /home/digoal/pgsql9.6/bin/postgres 0x3c762f d [.] ExecInitExpr
47. 0.36% 0.36% 0.00% 816 pgbench [kernel.kallsyms] 0xffffffff8119f950 k [k] do_select
48. 0.36% 0.00% 0.00% 4 postgres /home/digoal/pgsql9.6/bin/postgres 0x21e005 d [.] ReadBuffer_common.lto_priv.1215
49. 0.36% 0.00% 0.01% 28 postgres /home/digoal/pgsql9.6/bin/postgres 0x1e5e50 d [.] _bt_doinsert
50. 0.35% 0.00% 0.01% 16 postgres [vdso] 0x7fffb61ff8b0 ! [.] 0x7fffb61ff8b0
51. 0.35% 0.00% 0.00% 1 postgres /lib64/libc-2.12.so 0x9b6db B [.] gettimeofday
52. 0.35% 0.00% 0.35% 775 pgbench /home/digoal/pgsql9.6/lib/libpq.so.5.9 0x18057 d [.] pqParseInput3
53. 0.32% 0.00% 0.01% 18 postgres /home/digoal/pgsql9.6/bin/postgres 0x2f9979 d [.] expression_tree_walker
54. 0.31% 0.00% 0.01% 18 postgres /home/digoal/pgsql9.6/bin/postgres 0x306a5c d [.] socket_putmessage.lto_priv.1817
55. 0.31% 0.31% 0.00% 702 pgbench [kernel.kallsyms] 0xffffffff8148c9be k [k] tcp_sendmsg
56. 0.31% 0.00% 0.01% 17 postgres /home/digoal/pgsql9.6/bin/postgres 0x38891e d [.] ExecInitModifyTable
57. 0.31% 0.00% 0.00% 1 postgres [kernel.kallsyms] 0xffffffff81057ff7 k [k] update_curr
58. 0.31% 0.01% 0.00% 27 postgres [kernel.kallsyms] 0xffffffff81494ea6 k [k] tcp_ack
展开后的例子
1. - 1.39% 0.00% 0.01% 15 postgres /home/digoal/pgsql9.6/bin/postgres 0x524e3a d [.] AllocSetAlloc.lto_priv.1186
2. - AllocSetAlloc.lto_priv.1186
3. - 33.72% MemoryContextAlloc
4. - 27.57% MemoryContextCreate
5. - AllocSetContextCreate.constprop.955
6. - 43.77% StartTransaction.lto_priv.2177
7. start_xact_command.part.3
8. PostgresMain
9. ServerLoop
10. PostmasterMain
11. main
12. __libc_start_main
13. - 16.08% CreatePortal.constprop.203
14. PostgresMain
15. ServerLoop
16. PostmasterMain
17. main
18. __libc_start_main
19. - 14.06% MakePerTupleExprContext
20. ExecInsertIndexTuples.constprop.430
21. ExecModifyTable
22. ExecProcNode
23. standard_ExecutorRun
24. pgss_ExecutorRun
25. ProcessQuery
26. PortalRunMulti
27. PortalRun
28. PostgresMain
29. ServerLoop
30. PostmasterMain
31. main
32. __libc_start_main
33. - 13.35% CreateExecutorState
34. standard_ExecutorStart
35. pgss_ExecutorStart
36. ProcessQuery
37. PortalRunMulti
38. PortalRun
39. PostgresMain
40. ServerLoop
41. PostmasterMain
42. main
43. __libc_start_main
44. - 12.74% ExecInitResult
45. ExecInitNode
46. ExecInitModifyTable
47. ExecInitNode
48. standard_ExecutorStart
49. pgss_ExecutorStart
50. ProcessQuery
51. PortalRunMulti
52. PortalRun
53. PostgresMain
54. ServerLoop
55. ...................略................
如果要查看详情，建议使用—stdio模式，看得非常清晰
1. #perf report -v -n --showcpuutilization -g --stdio
3. # Events: 224K cycles
4. #
5. # Overhead Samples sys us Command Shared Object
6. (全局开销占比) (采样) (系统开销占比, 系统开销耗时us) (命令) (共享对象)
7. # ........ .......... .............. ............................................... ........................................................
8. #
9. 1.39% 0.00% 0.01% 15 postgres /home/digoal/pgsql9.6/bin/postgres 0x524e3a d [.] AllocSetAlloc.lto_priv.1186
10. |
11. --- AllocSetAlloc.lto_priv.1186
12. |
13. |--33.72%-- MemoryContextAlloc
14. | |
15. | |--27.57%-- MemoryContextCreate
16. | | AllocSetContextCreate.constprop.955
17. | | |
18. | | |--43.77%-- StartTransaction.lto_priv.2177
19. | | | start_xact_command.part.3
20. | | | PostgresMain
21. | | | ServerLoop
22. | | | PostmasterMain
23. | | | main
24. | | | __libc_start_main
25. | | |
26. | | |--16.08%-- CreatePortal.constprop.203
27. | | | PostgresMain
28. | | | ServerLoop
29. | | | PostmasterMain
30. | | | main
31. | | | __libc_start_main
32. | | |
33. | | |--14.06%-- MakePerTupleExprContext
34. | | | ExecInsertIndexTuples.constprop.430
35. | | | ExecModifyTable
36. | | | ExecProcNode
37. | | | standard_ExecutorRun
38. | | | pgss_ExecutorRun
39. | | | ProcessQuery
40. | | | PortalRunMulti
41. | | | PortalRun
42. | | | PostgresMain
43. | | | ServerLoop
44. | | | PostmasterMain
45. | | | main
46. | | | __libc_start_main
47. | | |
48. | | |--13.35%-- CreateExecutorState
49. | | | standard_ExecutorStart
50. | | | pgss_ExecutorStart
51. | | | ProcessQuery
52. | | | PortalRunMulti
53. | | | PortalRun
54. | | | PostgresMain
55. | | | ServerLoop
56. | | | PostmasterMain
57. | | | main
58. | | | __libc_start_main
59. ...............略.............................
perf report输出为本文，可以较为详细的进行代码瓶颈的分析。
6. 生成火焰图
如果你觉得perf report的文本输出不够直观，可以生成火焰图，就更加直观了。
生成火焰图需要将perf record的原始数据弄出来，然后绘图。
perf.data的原始数据perf script即可输出。
生成火焰图
1. # git clone https://github.com/brendangregg/FlameGraph # or download it from github
2. # mv perf.data FlameGraph/
3. # cd FlameGraph
4. # perf script | ./stackcollapse-perf.pl > out.perf-folded
5. # cat out.perf-folded | ./flamegraph.pl > perf-kernel.svg
svg可以使用浏览器打开，转换的图片如下
指定内核内置事件跟踪
perf top|stat|record 都支持通过-e指定事件进行跟踪。
支持的事件如下
1. perf list
3. cpu-cycles OR cycles [Hardware event]
4. stalled-cycles-frontend OR idle-cycles-frontend [Hardware event]
5. stalled-cycles-backend OR idle-cycles-backend [Hardware event]
6. instructions [Hardware event]
7. cache-references [Hardware event]
8. cache-misses [Hardware event]
9. branch-instructions OR branches [Hardware event]
10. branch-misses [Hardware event]
11. bus-cycles [Hardware event]
12. ......................略........................
事件支持写通配符，多个事件使用逗号隔开。
1. perf top -ag -e xfs:xfs_attr_list_sf,skb:kfree_skb,block:*,irq:softirq*
3. No CONFIG_PERF_EVENTS=y kernel support configured? 编译内核是需要配置CONFIG_PERF_EVENTS=y
1. perf stat -e xfs:xfs_attr_list_sf,skb:kfree_skb,block:*,irq:softirq* -- sleep 10
3. Performance counter stats for 'sleep 10':
5. 0 xfs:xfs_attr_list_sf
6. 0 skb:kfree_skb
7. 0 block:block_rq_abort
8. 0 block:block_rq_requeue
9. 0 block:block_rq_complete
10. 0 block:block_rq_insert
11. 0 block:block_rq_issue
12. 0 block:block_bio_bounce
13. 0 block:block_bio_complete
14. 0 block:block_bio_backmerge
15. 0 block:block_bio_frontmerge
16. 0 block:block_bio_queue
17. 0 block:block_getrq
18. 0 block:block_sleeprq
19. 0 block:block_plug
20. 0 block:block_unplug_timer
21. 0 block:block_unplug_io
22. 0 block:block_split
23. 0 block:block_remap
24. 0 block:block_rq_remap
25. 0 irq:softirq_entry
26. 0 irq:softirq_exit
27. 0 irq:softirq_raise
29. 10.001021697 seconds time elapsed
1. perf record -avg -e xfs:xfs_attr_list_sf,skb:kfree_skb,block:*,irq:softirq* -- sleep 10
3. perf report -vg --stdio
5. # Events: 17 skb:kfree_skb
6. #
7. # Overhead Command Shared Object Symbol
8. # ........ .............. ........................................... ................
9. #
10. 35.29% init [kernel.kallsyms] 0xffffffff8103c94b k [k] native_safe_halt
11. |
12. --- native_safe_halt
14. 11.76% syslog-ng /lib64/libpthread-2.12.so 0xea10 B [.] __recvmsg
15. |
16. --- __recvmsg
17. ...........略.......................
kernel source 动态跟踪
如果perf list列出的事件不能满足跟踪需求，你可以自行添加任意代码进行跟踪。
这个功能需要perf和内核的支持，详见《Linux 性能诊断(profiling) perf使用指南》
1. 1. Listing variables available for tcp_sendmsg():
3. # perf probe -V tcp_sendmsg
4. Available variables at tcp_sendmsg
5. @<tcp_sendmsg+0>
6. size_t size
7. struct kiocb* iocb
8. struct msghdr* msg
9. struct sock* sk
11. 2. Creating a probe for tcp_sendmsg() with the "size" variable:
13. # perf probe --add 'tcp_sendmsg size'
14. Added new event:
15. probe:tcp_sendmsg (on tcp_sendmsg with size)
17. 3. You can now use it in all perf tools, such as:
19. perf record -e probe:tcp_sendmsg -aR sleep 1
PostgreSQL source 动态跟踪
这个功能需要perf和软件编译时的支持，详见《Linux 性能诊断(profiling) perf使用指南》
1. sudo perf record -g dwarf -e block:block_rq_*,syscalls:sys_enter_write,syscalls:sys_enter_fsync -p 4495
1. sudo perf probe -x `which postgres` XLogFileInit
2. sudo perf probe -x `which postgres` XLogFileOpen
3. sudo perf probe -x `which postgres` XLogFlush
1. sudo perf probe -l
2. probe_postgres:XLogFileInit (on 0x000000000009a360)
3. probe_postgres:XLogFileOpen (on 0x000000000009a860)
4. probe_postgres:XLogFlush (on 0x00000000000a0670)
1. sudo perf record -g dwarf -u postgres -e probe_postgres:XLogFileInit,probe_postgres:XLogFileOpen,probe_postgres:XLogFlush
1. sudo perf stat -e probe_postgres:XLogFileInit,probe_postgres:XLogFileOpen,probe_postgres:XLogFlush -a -- /usr/pgsql-9.2/bin/pgbench -U postgres craig -c 2 -t 10000
2. starting vacuum...end.
3. transaction type: TPC-B (sort of)
4. scaling factor: 100
5. query mode: simple
6. number of clients: 2
7. number of threads: 1
8. number of transactions per client: 10000
9. number of transactions actually processed: 20000/20000
10. tps = 715.854663 (including connections establishing)
11. tps = 716.092133 (excluding connections establishing)
13. Performance counter stats for '/usr/pgsql-9.2/bin/pgbench -U postgres craig -c 2 -t 10000':
15. 64 probe_postgres:XLogFileInit [100.00%]
16. 0 probe_postgres:XLogFileOpen [100.00%]
17. 55,440 probe_postgres:XLogFlush
19. 27.987364469 seconds time elapsed
参考
-
http://blog.2ndquadrant.com/testing-postgresql-vagrant-computer/
-
http://blog.2ndquadrant.com/devops-testing-postgresql-with-vagrant-on-your-computer-part-two/
-
https://www.quora.com/How-do-I-compile-a-Linux-perf-tool-with-all-features-For-Linux-4-0-on-Ubuntu
