如何进行Fio随机读IOPS测试值可能偏大的原因分析，相信很多没有经验的人对此束手无策，为此本文总结了问题出现的原因和解决方法，通过这篇文章希望你能解决这个问题。
问题描述：
在使用fio进行虚拟机磁盘（Ceph的RBD，格式化为ext4文件系统）的IOPS测试时，发现randread比预估值高许多；
在使用相同参数进行randwrite测试之后，再进行randread时会出现此现象；
而使用dd构建测试文件后，再进行randread就不会出现这种情况，IOPS数值正常。
初步推测，可能fio的随机是伪随机，导致前后的randwrite和randread使用了相同的伪随机序列。文件系统在进行物理块分配时从前往后分配，在逻辑上随机的块实际上是顺序写入物理磁盘，最终的随机读实际上是顺序读，导致IO被磁盘调度器合并，实际IO次数变少，所以测试的IOPS偏大，为此进行详细分析测试。
打印Debug日志
$ fio -direct=1 -iodepth=128 -rw=randwrite -ioengine=libaio -bs=4k -size=1G -numjobs=1 -runtime=10 -group_reporting -filename=iotest -name=Rand_Write_Testing --debug=random > rand_write_offset.log
$ fio -direct=1 -iodepth=128 -rw=randread -ioengine=libaio -bs=4k -size=1G -numjobs=1 -runtime=10 -group_reporting -filename=iotest -name=Rand_Read_Testingg --debug=random > rand_read_offset.log
$ head -n30 rand_write_offset.log
fio: set debug option random
Rand_Write_Testing: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=128
fio-3.1
Starting 1 process
random 4057532 off rand 259043585
random 4057532 off rand 3179521932
random 4057532 off rand 3621444214
random 4057532 off rand 2018697059
random 4057532 off rand 1726199243
random 4057532 off rand 3608323581
random 4057532 off rand 1634212905
random 4057532 off rand 1518359867
random 4057532 off rand 3921331707
random 4057532 off rand 287004724
random 4057532 off rand 3673173177
random 4057532 off rand 2796675757
random 4057532 off rand 3988051731
random 4057532 off rand 1060357494
random 4057532 off rand 1685717462
random 4057532 off rand 2400737531
random 4057532 off rand 1891936796
random 4057532 off rand 3455447349
random 4057532 off rand 1553547805
random 4057532 off rand 2660809810
random 4057532 off rand 17263379
random 4057532 off rand 1823528783
random 4057532 off rand 1355450167
random 4057532 off rand 2956359995
random 4057532 off rand 3392712188
random 4057532 off rand 4240594610
$ $ head -n30 rand_read_offset.log
fio: set debug option random
Rand_Read_Testingg: (g=0): rw=randread, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=128
fio-3.1
Starting 1 process
random 4057831 off rand 259043585
random 4057831 off rand 3179521932
random 4057831 off rand 3621444214
random 4057831 off rand 2018697059
random 4057831 off rand 1726199243
random 4057831 off rand 3608323581
random 4057831 off rand 1634212905
random 4057831 off rand 1518359867
random 4057831 off rand 3921331707
random 4057831 off rand 287004724
random 4057831 off rand 3673173177
random 4057831 off rand 2796675757
random 4057831 off rand 3988051731
random 4057831 off rand 1060357494
random 4057831 off rand 1685717462
random 4057831 off rand 2400737531
random 4057831 off rand 1891936796
random 4057831 off rand 3455447349
random 4057831 off rand 1553547805
random 4057831 off rand 2660809810
random 4057831 off rand 17263379
random 4057831 off rand 1823528783
random 4057831 off rand 1355450167
random 4057831 off rand 2956359995
random 4057831 off rand 3392712188
random 4057831 off rand 4240594610
获取Fio源码
$ git clone https://github.com/axboe/fio.git
$ cd fio
$ git branch -av
* master ee636f3 libaio: switch to newer libaio polled IO API
remotes/origin/HEAD -> origin/master
remotes/origin/latency-probe fcd4e74 target: fixes
remotes/origin/master ee636f3 libaio: switch to newer libaio polled IO API
分析debug选项
$ grep -rHn \"debug\"
init.c:176: .name = (char *) "debug",
$ grep -rHn \"random\" -A5 init.c
init.c:2260: { .name = "random",
init.c-2261- .help = "Random generation logging",
init.c-2262- .shift = FD_RANDOM,
init.c-2263- },
init.c-2264- { .name = "parse",
init.c-2265- .help = "Parser logging",$ grep -rHn FD_RANDOM
debug.h:13: FD_RANDOM,
init.c:2262: .shift = FD_RANDOM,
io_u.c:98: dprint(FD_RANDOM, "off rand %llu\n", (unsigned long long) r);
io_u.c:124: dprint(FD_RANDOM, "get_next_rand_offset: offset %llu busy\n",
$ grep -rHn FD_RANDOM io_u.c -C12
io_u.c-86-
io_u.c-87-static int __get_next_rand_offset(struct thread_data *td, struct fio_file *f,
io_u.c-88- enum fio_ddir ddir, uint64_t *b,
io_u.c-89- uint64_t lastb)
io_u.c-90-{
io_u.c-91- uint64_t r;
io_u.c-92-
io_u.c-93- if (td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE ||
io_u.c-94- td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE64) {
io_u.c-95-
io_u.c-96- r = __rand(&td->random_state);
io_u.c-97-
io_u.c:98: dprint(FD_RANDOM, "off rand %llu\n", (unsigned long long) r);
io_u.c-99-
io_u.c-100- *b = lastb * (r / (rand_max(&td->random_state) + 1.0));
io_u.c-101- } else {
io_u.c-102- uint64_t off = 0;
io_u.c-103-
io_u.c-104- assert(fio_file_lfsr(f));
io_u.c-105-
io_u.c-106- if (lfsr_next(&f->lfsr, &off))
io_u.c-107- return 1;
io_u.c-108-
io_u.c-109- *b = off;
io_u.c-110- }
--
io_u.c-112- /*
io_u.c-113- * if we are not maintaining a random map, we are done.
io_u.c-114- */
io_u.c-115- if (!file_randommap(td, f))
io_u.c-116- goto ret;
io_u.c-117-
io_u.c-118- /*
io_u.c-119- * calculate map offset and check if it's free
io_u.c-120- */
io_u.c-121- if (random_map_free(f, *b))
io_u.c-122- goto ret;
io_u.c-123-
io_u.c:124: dprint(FD_RANDOM, "get_next_rand_offset: offset %llu busy\n",
io_u.c-125- (unsigned long long) *b);
io_u.c-126-
io_u.c-127- *b = axmap_next_free(f->io_axmap, *b);
io_u.c-128- if (*b == (uint64_t) -1ULL)
io_u.c-129- return 1;
io_u.c-130-ret:
io_u.c-131- return 0;
io_u.c-132-}
io_u.c-133-
io_u.c-134-static int __get_next_rand_offset_zipf(struct thread_data *td,
io_u.c-135- struct fio_file *f, enum fio_ddir ddir,
io_u.c-136- uint64_t *b)分析dprint函数
$ grep -rHn " dprint"
debug.h:62:#define dprint(type, str, args...) \
debug.h:71:static inline void dprint(int type, const char *str, ...)
gettime.c:320: dprint(FD_TIME, "tmp=%llu, sft=%u\n", tmp, sft);
io_u.h:153:static inline void dprint_io_u(struct io_u *io_u, const char *p)
io_u.h:170:#define dprint_io_u(io_u, p)
t/time-test.c:88:#define dprintf(...) if (DEBUG) { printf(__VA_ARGS__); }$ grep -rHn " dprint" -C7 debug.h
debug.h-55-};
debug.h-56-extern const struct debug_level debug_levels[];
debug.h-57-
debug.h-58-extern unsigned long fio_debug;
debug.h-59-
debug.h-60-void __dprint(int type, const char *str, ...) __attribute__((format (printf, 2, 3)));
debug.h-61-
debug.h:62:#define dprint(type, str, args...) \
debug.h-63- do { \
debug.h-64- if (((1 << type) & fio_debug) == 0) \
debug.h-65- break; \
debug.h-66- __dprint((type), (str), ##args); \
debug.h-67- } while (0) \
debug.h-68-
debug.h-69-#else
debug.h-70-
debug.h:71:static inline void dprint(int type, const char *str, ...)
debug.h-72-{
debug.h-73-}
debug.h-74-#endif
debug.h-75-
debug.h-76-#endif$ grep -rHn " __dprint" -C10
debug.c-1-#include <assert.h>
debug.c-2-#include <stdarg.h>
debug.c-3-
debug.c-4-#include "debug.h"
debug.c-5-#include "log.h"
debug.c-6-
debug.c-7-#ifdef FIO_INC_DEBUG
debug.c:8:void __dprint(int type, const char *str, ...)
debug.c-9-{
debug.c-10- va_list args;
debug.c-11-
debug.c-12- assert(type < FD_DEBUG_MAX);
debug.c-13-
debug.c-14- va_start(args, str);
debug.c-15- log_prevalist(type, str, args);
debug.c-16- va_end(args);
debug.c-17-}
debug.c-18-#endif
--
debug.h-50-struct debug_level {
debug.h-51- const char *name;
debug.h-52- const char *help;
debug.h-53- unsigned long shift;
debug.h-54- unsigned int jobno;
debug.h-55-};
debug.h-56-extern const struct debug_level debug_levels[];
debug.h-57-
debug.h-58-extern unsigned long fio_debug;
debug.h-59-
debug.h:60:void __dprint(int type, const char *str, ...) __attribute__((format (printf, 2, 3)));
debug.h-61-
debug.h-62-#define dprint(type, str, args...) \
debug.h-63- do { \
debug.h-64- if (((1 << type) & fio_debug) == 0) \
debug.h-65- break; \
debug.h-66- __dprint((type), (str), ##args); \
debug.h-67- } while (0) \
debug.h-68-
debug.h-69-#else
debug.h-70-
--
t/debug.c-1-#include <stdio.h>
t/debug.c-2-
t/debug.c-3-FILE *f_err;
t/debug.c-4-struct timespec *fio_ts = NULL;
t/debug.c-5-unsigned long fio_debug = 0;
t/debug.c-6-
t/debug.c:7:void __dprint(int type, const char *str, ...)
t/debug.c-8-{
t/debug.c-9-}
t/debug.c-10-
t/debug.c-11-void debug_init(void)
t/debug.c-12-{
t/debug.c-13- f_err = stderr;
t/debug.c-14-}/* add prefix for the specified type in front of the valist */
void log_prevalist(int type, const char *fmt, va_list args)
{
char *buf1, *buf2;
int len;
pid_t pid;
pid = gettid();
if (fio_debug_jobp && *fio_debug_jobp != -1U
&& pid != *fio_debug_jobp)
return;
len = vasprintf(&buf1, fmt, args);
if (len < 0)
return;
len = asprintf(&buf2, "%-8s %-5u %s", debug_levels[type].name,
(int) pid, buf1);
free(buf1);
if (len < 0)
return;
len = log_info_buf(buf2, len);
free(buf2);
}$ grep -rHn debug_levels
debug.h:56:extern const struct debug_level debug_levels[];
gfio.c:1187: buttons[i] = gtk_check_button_new_with_label(debug_levels[i].name);
gfio.c:1188: gtk_widget_set_tooltip_text(buttons[i], debug_levels[i].help);
init.c:2144: const struct debug_level *dl = &debug_levels[0];
init.c:2235:const struct debug_level debug_levels[] = {
init.c:2327: for (i = 0; debug_levels[i].name; i++) {
init.c:2328: dl = &debug_levels[i];
init.c:2344: for (i = 0; debug_levels[i].name; i++) {
init.c:2345: dl = &debug_levels[i];
log.c:59: len = asprintf(&buf2, "%-8s %-5u %s", debug_levels[type].name,#ifdef FIO_INC_DEBUG
const struct debug_level debug_levels[] = {
{ .name = "process",
.help = "Process creation/exit logging",
.shift = FD_PROCESS,
},
{ .name = "file",
.help = "File related action logging",
.shift = FD_FILE,
},
{ .name = "io",
.help = "IO and IO engine action logging (offsets, queue, completions, etc)",
.shift = FD_IO,
},
{ .name = "mem",
.help = "Memory allocation/freeing logging",
.shift = FD_MEM,
},
{ .name = "blktrace",
.help = "blktrace action logging",
.shift = FD_BLKTRACE,
},
{ .name = "verify",
.help = "IO verification action logging",
.shift = FD_VERIFY,
},
{ .name = "random",
.help = "Random generation logging",
.shift = FD_RANDOM,
},
{ .name = "parse",
.help = "Parser logging",
.shift = FD_PARSE,
},
{ .name = "diskutil",
.help = "Disk utility logging actions",
.shift = FD_DISKUTIL,
},
{ .name = "job",
.help = "Logging related to creating/destroying jobs",
.shift = FD_JOB,
},
{ .name = "mutex",
.help = "Mutex logging",
.shift = FD_MUTEX
},
{ .name = "profile",
.help = "Logging related to profiles",
.shift = FD_PROFILE,
},
{ .name = "time",
.help = "Logging related to time keeping functions",
.shift = FD_TIME,
},
{ .name = "net",
.help = "Network logging",
.shift = FD_NET,
},
{ .name = "rate",
.help = "Rate logging",
.shift = FD_RATE,
},
{ .name = "compress",
.help = "Log compression logging",
.shift = FD_COMPRESS,
},
{ .name = "steadystate",
.help = "Steady state detection logging",
.shift = FD_STEADYSTATE,
},
{ .name = "helperthread",
.help = "Helper thread logging",
.shift = FD_HELPERTHREAD,
},
{ .name = "zbd",
.help = "Zoned Block Device logging",
.shift = FD_ZBD,
},
{ .name = NULL, },
};
static int set_debug(const char *string)
{
const struct debug_level *dl;
char *p = (char *) string;
char *opt;
int i;
if (!string)
return 0;
if (!strcmp(string, "?") || !strcmp(string, "help")) {
log_info("fio: dumping debug options:");
for (i = 0; debug_levels[i].name; i++) {
dl = &debug_levels[i];
log_info("%s,", dl->name);
}
log_info("all\n");
return 1;
}
while ((opt = strsep(&p, ",")) != NULL) {
int found = 0;
if (!strncmp(opt, "all", 3)) {
log_info("fio: set all debug options\n");
fio_debug = ~0UL;
continue;
}
for (i = 0; debug_levels[i].name; i++) {
dl = &debug_levels[i];
found = !strncmp(opt, dl->name, strlen(dl->name));
if (!found)
continue;
if (dl->shift == FD_JOB) {
opt = strchr(opt, ':');
if (!opt) {
log_err("fio: missing job number\n");
break;
}
opt++;
fio_debug_jobno = atoi(opt);
log_info("fio: set debug jobno %d\n",
fio_debug_jobno);
} else {
log_info("fio: set debug option %s\n", opt);
fio_debug |= (1UL << dl->shift);
}
break;
}
if (!found)
log_err("fio: debug mask %s not found\n", opt);
}
return 0;
}
#else
static int set_debug(const char *string)
{
log_err("fio: debug tracing not included in build\n");
return 1;
}
#endif分析randwrite选项
$ grep -rHn \"randwrite\" -C5
io_ddir.h-62-}
io_ddir.h-63-
io_ddir.h-64-static inline const char *ddir_str(enum td_ddir ddir)
io_ddir.h-65-{
io_ddir.h-66- static const char *__str[] = { NULL, "read", "write", "rw", "rand",
io_ddir.h:67: "randread", "randwrite", "randrw",
io_ddir.h-68- "trim", NULL, "trimwrite", NULL, "randtrim" };
io_ddir.h-69-
io_ddir.h-70- return __str[ddir];
io_ddir.h-71-}
io_ddir.h-72-
--
options.c-1690- },
options.c-1691- { .ival = "randread",
options.c-1692- .oval = TD_DDIR_RANDREAD,
options.c-1693- .help = "Random read",
options.c-1694- },
options.c:1695: { .ival = "randwrite",
options.c-1696- .oval = TD_DDIR_RANDWRITE,
options.c-1697- .help = "Random write",
options.c-1698- },
options.c-1699- { .ival = "randtrim",
options.c-1700- .oval = TD_DDIR_RANDTRIM,
--
profiles/act.c-182-
profiles/act.c-183- if (act_add_opt("name=act-%s-%s", reads ? "read" : "write", dev))
profiles/act.c-184- return 1;
profiles/act.c-185- if (act_add_opt("filename=%s", dev))
profiles/act.c-186- return 1;
profiles/act.c:187: if (act_add_opt("rw=%s", reads ? "randread" : "randwrite"))
profiles/act.c-188- return 1;
profiles/act.c-189- if (reads) {
profiles/act.c-190- int rload = ao->load * R_LOAD / ao->threads_per_queue;
profiles/act.c-191-
profiles/act.c-192- if (act_add_opt("numjobs=%u", ao->threads_per_queue))
--
t/sgunmap-test.py-116-
t/sgunmap-test.py-117-
t/sgunmap-test.py-118-def runalltests(args, qd, batch):
t/sgunmap-test.py-119- block = False
t/sgunmap-test.py-120- for dev in [args.chardev, args.blockdev]:
t/sgunmap-test.py:121: for rw in ["randread", "randwrite", "randtrim"]:
t/sgunmap-test.py-122- parameters = ["--name=test",
t/sgunmap-test.py-123- "--time_based",
t/sgunmap-test.py-124- "--runtime=30s",
t/sgunmap-test.py-125- "--output-format=json",
t/sgunmap-test.py-126- "--ioengine=sg",$ grep -rHn TD_DDIR_RANDWRITE
io_ddir.h:38: TD_DDIR_RANDWRITE = TD_DDIR_WRITE | TD_DDIR_RAND,
options.c:1696: .oval = TD_DDIR_RANDWRITE,
$ grep -rHn TD_DDIR_RAND -C3
io_ddir.h-31-enum td_ddir {
io_ddir.h-32- TD_DDIR_READ = 1 << 0,
io_ddir.h-33- TD_DDIR_WRITE = 1 << 1,
io_ddir.h:34: TD_DDIR_RAND = 1 << 2,
io_ddir.h-35- TD_DDIR_TRIM = 1 << 3,
io_ddir.h-36- TD_DDIR_RW = TD_DDIR_READ | TD_DDIR_WRITE,
io_ddir.h:37: TD_DDIR_RANDREAD = TD_DDIR_READ | TD_DDIR_RAND,
io_ddir.h:38: TD_DDIR_RANDWRITE = TD_DDIR_WRITE | TD_DDIR_RAND,
io_ddir.h:39: TD_DDIR_RANDRW = TD_DDIR_RW | TD_DDIR_RAND,
io_ddir.h:40: TD_DDIR_RANDTRIM = TD_DDIR_TRIM | TD_DDIR_RAND,
io_ddir.h-41- TD_DDIR_TRIMWRITE = TD_DDIR_TRIM | TD_DDIR_WRITE,
io_ddir.h-42-};
io_ddir.h-43-
--
io_ddir.h-45-#define td_write(td) ((td)->o.td_ddir & TD_DDIR_WRITE)
io_ddir.h-46-#define td_trim(td) ((td)->o.td_ddir & TD_DDIR_TRIM)
io_ddir.h-47-#define td_rw(td) (((td)->o.td_ddir & TD_DDIR_RW) == TD_DDIR_RW)
io_ddir.h:48:#define td_random(td) ((td)->o.td_ddir & TD_DDIR_RAND)
io_ddir.h-49-#define file_randommap(td, f) (!(td)->o.norandommap && fio_file_axmap((f)))
io_ddir.h-50-#define td_trimwrite(td) (((td)->o.td_ddir & TD_DDIR_TRIMWRITE) \
io_ddir.h-51- == TD_DDIR_TRIMWRITE)
--
options.c-1689- .help = "Sequential trim",
options.c-1690- },
options.c-1691- { .ival = "randread",
options.c:1692: .oval = TD_DDIR_RANDREAD,
options.c-1693- .help = "Random read",
options.c-1694- },
options.c-1695- { .ival = "randwrite",
options.c:1696: .oval = TD_DDIR_RANDWRITE,
options.c-1697- .help = "Random write",
options.c-1698- },
options.c-1699- { .ival = "randtrim",
options.c:1700: .oval = TD_DDIR_RANDTRIM,
options.c-1701- .help = "Random trim",
options.c-1702- },
options.c-1703- { .ival = "rw",
--
options.c-1709- .help = "Sequential read and write mix",
options.c-1710- },
options.c-1711- { .ival = "randrw",
options.c:1712: .oval = TD_DDIR_RANDRW,
options.c-1713- .help = "Random read and write mix"
options.c-1714- },
options.c-1715- { .ival = "trimwrite",$ grep -rHn td_random -C5 io_u.c
io_u.c-416- assert(ddir_rw(ddir));
io_u.c-417-
io_u.c-418- b = offset = -1ULL;
io_u.c-419-
io_u.c-420- if (rw_seq) {
io_u.c:421: if (td_random(td)) {
io_u.c-422- if (should_do_random(td, ddir)) {
io_u.c-423- ret = get_next_rand_block(td, f, ddir, &b);
io_u.c-424- *is_random = true;
io_u.c-425- } else {
io_u.c-426- *is_random = false;
--
io_u.c-934- }
io_u.c-935-
io_u.c-936- /*
io_u.c-937- * mark entry before potentially trimming io_u
io_u.c-938- */
io_u.c:939: if (td_random(td) && file_randommap(td, io_u->file))
io_u.c-940- io_u->buflen = mark_random_map(td, io_u, offset, io_u->buflen);
io_u.c-941-
io_u.c-942-out:
io_u.c-943- dprint_io_u(io_u, "fill");
io_u.c-944- td->zone_bytes += io_u->buflen;static int get_next_rand_block(struct thread_data *td, struct fio_file *f,
enum fio_ddir ddir, uint64_t *b)
{
if (!get_next_rand_offset(td, f, ddir, b))
return 0;
if (td->o.time_based ||
(td->o.file_service_type & __FIO_FSERVICE_NONUNIFORM)) {
fio_file_reset(td, f);
loop_cache_invalidate(td, f);
if (!get_next_rand_offset(td, f, ddir, b))
return 0;
}
dprint(FD_IO, "%s: rand offset failed, last=%llu, size=%llu\n",
f->file_name, (unsigned long long) f->last_pos[ddir],
(unsigned long long) f->real_file_size);
return 1;
}
static int get_next_rand_offset(struct thread_data *td, struct fio_file *f,
enum fio_ddir ddir, uint64_t *b)
{
if (td->o.random_distribution == FIO_RAND_DIST_RANDOM) {
uint64_t lastb;
lastb = last_block(td, f, ddir);
if (!lastb)
return 1;
return __get_next_rand_offset(td, f, ddir, b, lastb);
} else if (td->o.random_distribution == FIO_RAND_DIST_ZIPF)
return __get_next_rand_offset_zipf(td, f, ddir, b);
else if (td->o.random_distribution == FIO_RAND_DIST_PARETO)
return __get_next_rand_offset_pareto(td, f, ddir, b);
else if (td->o.random_distribution == FIO_RAND_DIST_GAUSS)
return __get_next_rand_offset_gauss(td, f, ddir, b);
else if (td->o.random_distribution == FIO_RAND_DIST_ZONED)
return __get_next_rand_offset_zoned(td, f, ddir, b);
else if (td->o.random_distribution == FIO_RAND_DIST_ZONED_ABS)
return __get_next_rand_offset_zoned_abs(td, f, ddir, b);
log_err("fio: unknown random distribution: %d\n", td->o.random_distribution);
return 1;
}
static int __get_next_rand_offset(struct thread_data *td, struct fio_file *f,
enum fio_ddir ddir, uint64_t *b,
uint64_t lastb)
{
uint64_t r;
if (td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE ||
td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE64) {
r = __rand(&td->random_state);
dprint(FD_RANDOM, "off rand %llu\n", (unsigned long long) r);
*b = lastb * (r / (rand_max(&td->random_state) + 1.0));
} else {
uint64_t off = 0;
assert(fio_file_lfsr(f));
if (lfsr_next(&f->lfsr, &off))
return 1;
*b = off;
}
/*
* if we are not maintaining a random map, we are done.
*/
if (!file_randommap(td, f))
goto ret;
/*
* calculate map offset and check if it's free
*/
if (random_map_free(f, *b))
goto ret;
dprint(FD_RANDOM, "get_next_rand_offset: offset %llu busy\n",
(unsigned long long) *b);
*b = axmap_next_free(f->io_axmap, *b);
if (*b == (uint64_t) -1ULL)
return 1;
ret:
return 0;
}分析随机数计算
$ grep -rHn " __rand("
backend.c:1012: io_u->rand_seed = __rand(&td->verify_state);
backend.c:1014: io_u->rand_seed *= __rand(&td->verify_state);
engines/rdma.c:715: index = __rand(&rd->rand_state) % rd->rmt_nr;
engines/rdma.c:725: index = __rand(&rd->rand_state) % rd->rmt_nr;
filesetup.c:337: r = __rand(&td->file_size_state);
io_u.c:96: r = __rand(&td->random_state);
io_u.c:548: r = __rand(&td->bsrange_state[ddir]);
io_u.c:1165: r = __rand(&td->next_file_state);
lib/gauss.c:16: r = __rand(&gs->r);
lib/gauss.c:28: sum += __rand(&gs->r) % (gs->nranges + 1);
lib/rand.c:128: unsigned long r = __rand(fs);
lib/rand.c:131: r *= (unsigned long) __rand(fs);
lib/rand.c:190: unsigned long r = __rand(fs);
lib/rand.c:193: r *= (unsigned long) __rand(fs);
lib/rand.h:96:static inline uint64_t __rand(struct frand_state *state)
lib/zipf.c:32: zs->rand_off = __rand(&zs->rand);
lib/zipf.c:55: rand_uni = (double) __rand(&zs->rand) / (double) FRAND32_MAX;
lib/zipf.c:82: double rand = (double) __rand(&zs->rand) / (double) FRAND32_MAX;
trim.c:77: r = __rand(&td->trim_state);
verify.c:1370: io_u->rand_seed = __rand(&td->verify_state);
verify.c:1372: io_u->rand_seed *= __rand(&td->verify_state);struct taus88_state {
unsigned int s1, s2, s3;
};
struct taus258_state {
uint64_t s1, s2, s3, s4, s5;
};
struct frand_state {
unsigned int use64;
union {
struct taus88_state state32;
struct taus258_state state64;
};
};
static inline unsigned int __rand32(struct taus88_state *state)
{
#define TAUSWORTHE(s,a,b,c,d) ((s&c)<<d) ^ (((s <<a) ^ s)>>b)
state->s1 = TAUSWORTHE(state->s1, 13, 19, 4294967294UL, 12);
state->s2 = TAUSWORTHE(state->s2, 2, 25, 4294967288UL, 4);
state->s3 = TAUSWORTHE(state->s3, 3, 11, 4294967280UL, 17);
return (state->s1 ^ state->s2 ^ state->s3);
}
static inline uint64_t __rand64(struct taus258_state *state)
{
uint64_t xval;
xval = ((state->s1 << 1) ^ state->s1) >> 53;
state->s1 = ((state->s1 & 18446744073709551614ULL) << 10) ^ xval;
xval = ((state->s2 << 24) ^ state->s2) >> 50;
state->s2 = ((state->s2 & 18446744073709551104ULL) << 5) ^ xval;
xval = ((state->s3 << 3) ^ state->s3) >> 23;
state->s3 = ((state->s3 & 18446744073709547520ULL) << 29) ^ xval;
xval = ((state->s4 << 5) ^ state->s4) >> 24;
state->s4 = ((state->s4 & 18446744073709420544ULL) << 23) ^ xval;
xval = ((state->s5 << 3) ^ state->s5) >> 33;
state->s5 = ((state->s5 & 18446744073701163008ULL) << 8) ^ xval;
return (state->s1 ^ state->s2 ^ state->s3 ^ state->s4 ^ state->s5);
}
static inline uint64_t __rand(struct frand_state *state)
{
if (state->use64)
return __rand64(&state->state64);
else
return __rand32(&state->state32);
}$ grep -rHn " rand_max("
filesetup.c:336: frand_max = rand_max(&td->file_size_state);
io_u.c:546: frand_max = rand_max(&td->bsrange_state[ddir]);
io_u.c:1162: uint64_t frand_max = rand_max(&td->next_file_state);
lib/rand.h:27:static inline uint64_t rand_max(struct frand_state *state)
trim.c:76: frand_max = rand_max(&td->trim_state);#define FRAND32_MAX (-1U)
#define FRAND64_MAX (-1ULL)
static inline uint64_t rand_max(struct frand_state *state)
{
if (state->use64)
return FRAND64_MAX;
else
return FRAND32_MAX;
}分析种子初始化
$ grep -rHn random_state
fio.h:357: struct frand_state random_state;
init.c:1056: init_rand_seed(&td->random_state, td->rand_seeds[FIO_RAND_BLOCK_OFF], use64);
io_u.c:96: r = __rand(&td->random_state);
io_u.c:100: *b = lastb * (r / (rand_max(&td->random_state) + 1.0));
verify.c:1623: if (td->random_state.use64) {
verify.c:1624: s->rand.state64.s[0] = cpu_to_le64(td->random_state.state64.s1);
verify.c:1625: s->rand.state64.s[1] = cpu_to_le64(td->random_state.state64.s2);
verify.c:1626: s->rand.state64.s[2] = cpu_to_le64(td->random_state.state64.s3);
verify.c:1627: s->rand.state64.s[3] = cpu_to_le64(td->random_state.state64.s4);
verify.c:1628: s->rand.state64.s[4] = cpu_to_le64(td->random_state.state64.s5);
verify.c:1632: s->rand.state32.s[0] = cpu_to_le32(td->random_state.state32.s1);
verify.c:1633: s->rand.state32.s[1] = cpu_to_le32(td->random_state.state32.s2);
verify.c:1634: s->rand.state32.s[2] = cpu_to_le32(td->random_state.state32.s3);// lib/rand.h
static inline uint64_t __rand64(struct taus258_state *state)
{
uint64_t xval;
xval = ((state->s1 << 1) ^ state->s1) >> 53;
state->s1 = ((state->s1 & 18446744073709551614ULL) << 10) ^ xval;
xval = ((state->s2 << 24) ^ state->s2) >> 50;
state->s2 = ((state->s2 & 18446744073709551104ULL) << 5) ^ xval;
xval = ((state->s3 << 3) ^ state->s3) >> 23;
state->s3 = ((state->s3 & 18446744073709547520ULL) << 29) ^ xval;
xval = ((state->s4 << 5) ^ state->s4) >> 24;
state->s4 = ((state->s4 & 18446744073709420544ULL) << 23) ^ xval;
xval = ((state->s5 << 3) ^ state->s5) >> 33;
state->s5 = ((state->s5 & 18446744073701163008ULL) << 8) ^ xval;
return (state->s1 ^ state->s2 ^ state->s3 ^ state->s4 ^ state->s5);
}
// lib/rand.c
static inline uint64_t __seed(uint64_t x, uint64_t m)
{
return (x < m) ? x + m : x;
}
static void __init_rand32(struct taus88_state *state, unsigned int seed)
{
int cranks = 6;
#define LCG(x, seed) ((x) * 69069 ^ (seed))
state->s1 = __seed(LCG((2^31) + (2^17) + (2^7), seed), 1);
state->s2 = __seed(LCG(state->s1, seed), 7);
state->s3 = __seed(LCG(state->s2, seed), 15);
while (cranks--)
__rand32(state);
}
static void __init_rand64(struct taus258_state *state, uint64_t seed)
{
int cranks = 6;
#define LCG64(x, seed) ((x) * 6906969069ULL ^ (seed))
state->s1 = __seed(LCG64((2^31) + (2^17) + (2^7), seed), 1);
state->s2 = __seed(LCG64(state->s1, seed), 7);
state->s3 = __seed(LCG64(state->s2, seed), 15);
state->s4 = __seed(LCG64(state->s3, seed), 33);
state->s5 = __seed(LCG64(state->s4, seed), 49);
while (cranks--)
__rand64(state);
}
void init_rand(struct frand_state *state, bool use64)
{
state->use64 = use64;
if (!use64)
__init_rand32(&state->state32, 1);
else
__init_rand64(&state->state64, 1);
}
void init_rand_seed(struct frand_state *state, unsigned int seed, bool use64)
{
state->use64 = use64;
if (!use64)
__init_rand32(&state->state32, seed);
else
__init_rand64(&state->state64, seed);
}分析种子来源
$ grep -rHn "rand_seeds\[FIO_RAND_BLOCK_OFF\]"
filesetup.c:1292: seed = td->rand_seeds[FIO_RAND_BLOCK_OFF];
filesetup.c:1882: lfsr_reset(&f->lfsr, td->rand_seeds[FIO_RAND_BLOCK_OFF]);
init.c:1054: td->rand_seeds[FIO_RAND_BLOCK_OFF] = FIO_RANDSEED * td->thread_number;
init.c:1056: init_rand_seed(&td->random_state, td->rand_seeds[FIO_RAND_BLOCK_OFF], use64);
// init.c
static void td_fill_rand_seeds_internal(struct thread_data *td, bool use64)
{
unsigned int read_seed = td->rand_seeds[FIO_RAND_BS_OFF];
unsigned int write_seed = td->rand_seeds[FIO_RAND_BS1_OFF];
unsigned int trim_seed = td->rand_seeds[FIO_RAND_BS2_OFF];
int i;
/*
* trimwrite is special in that we need to generate the same
* offsets to get the "write after trim" effect. If we are
* using bssplit to set buffer length distributions, ensure that
* we seed the trim and write generators identically. Ditto for
* verify, read and writes must have the same seed, if we are doing
* read verify.
*/
if (td->o.verify != VERIFY_NONE)
write_seed = read_seed;
if (td_trimwrite(td))
trim_seed = write_seed;
init_rand_seed(&td->bsrange_state[DDIR_READ], read_seed, use64);
init_rand_seed(&td->bsrange_state[DDIR_WRITE], write_seed, use64);
init_rand_seed(&td->bsrange_state[DDIR_TRIM], trim_seed, use64);
td_fill_verify_state_seed(td);
init_rand_seed(&td->rwmix_state, td->rand_seeds[FIO_RAND_MIX_OFF], false);
if (td->o.file_service_type == FIO_FSERVICE_RANDOM)
init_rand_seed(&td->next_file_state, td->rand_seeds[FIO_RAND_FILE_OFF], use64);
else if (td->o.file_service_type & __FIO_FSERVICE_NONUNIFORM)
init_rand_file_service(td);
init_rand_seed(&td->file_size_state, td->rand_seeds[FIO_RAND_FILE_SIZE_OFF], use64);
init_rand_seed(&td->trim_state, td->rand_seeds[FIO_RAND_TRIM_OFF], use64);
init_rand_seed(&td->delay_state, td->rand_seeds[FIO_RAND_START_DELAY], use64);
init_rand_seed(&td->poisson_state[0], td->rand_seeds[FIO_RAND_POISSON_OFF], 0);
init_rand_seed(&td->poisson_state[1], td->rand_seeds[FIO_RAND_POISSON2_OFF], 0);
init_rand_seed(&td->poisson_state[2], td->rand_seeds[FIO_RAND_POISSON3_OFF], 0);
init_rand_seed(&td->dedupe_state, td->rand_seeds[FIO_DEDUPE_OFF], false);
init_rand_seed(&td->zone_state, td->rand_seeds[FIO_RAND_ZONE_OFF], false);
if (!td_random(td))
return;
if (td->o.rand_repeatable)
td->rand_seeds[FIO_RAND_BLOCK_OFF] = FIO_RANDSEED * td->thread_number;
init_rand_seed(&td->random_state, td->rand_seeds[FIO_RAND_BLOCK_OFF], use64);
for (i = 0; i < DDIR_RWDIR_CNT; i++) {
struct frand_state *s = &td->seq_rand_state[i];
init_rand_seed(s, td->rand_seeds[FIO_RAND_SEQ_RAND_READ_OFF], false);
}
}$ grep -rHn thread_number
HOWTO:1245: * thread_number`, where the thread number is a counter that starts at 0 and
backend.c:64:unsigned int thread_number = 0;
backend.c:1624: ret = fio_cpus_split(&o->cpumask, td->thread_number - 1);
backend.c:1899: verify_save_state(td->thread_number);
backend.c:2131: td->thread_number - 1, &data);
backend.c:2248: todo = thread_number;
backend.c:2254: print_status_init(td->thread_number - 1);
backend.c:2488: if (!thread_number)
client.c:923: pdu.thread_number = cpu_to_le32(client->thread_number);
client.c:948: dst->thread_number = le32_to_cpu(src->thread_number);
client.c:1078: if (client->opt_lists && p->ts.thread_number <= client->jobs)
client.c:1079: opt_list = &client->opt_lists[p->ts.thread_number - 1];
client.c:1095: client_ts.thread_number = p->ts.thread_number;
client.c:1653: ret->thread_number = le32_to_cpu(ret->thread_number);
client.c:1832: client->thread_number = le32_to_cpu(pdu->thread_number);
client.h:60: uint32_t thread_number;
eta.c:41: char c = __run_str[td->thread_number - 1];
eta.c:118: __run_str[td->thread_number - 1] = c;
eta.c:411: eta_secs = malloc(thread_number * sizeof(uint64_t));
eta.c:412: memset(eta_secs, 0, thread_number * sizeof(uint64_t));
eta.c:530: je->nr_threads = thread_number;
eta.c:704: if (!thread_number)
filesetup.c:1203: seed = jhash(f->file_name, strlen(f->file_name), 0) * td->thread_number;
fio.1:1012:* thread_number', where the thread number is a counter that starts at 0 and
fio.h:183: unsigned int thread_number;
fio.h:509:extern unsigned int thread_number;
fio.h:701: for ((i) = 0, (td) = &threads[0]; (i) < (int) thread_number; (i)++, (td)++)
gclient.c:299: client_ts.thread_number = p->ts.thread_number;
gclient.c:578: p->thread_number = le32_to_cpu(p->thread_number);
init.c:480: if (thread_number >= max_jobs) {
init.c:486: td = &threads[thread_number++];
init.c:505: td->thread_number = thread_number;
init.c:536: memset(&threads[td->thread_number - 1], 0, sizeof(*td));
init.c:537: thread_number--;
init.c:1054: td->rand_seeds[FIO_RAND_BLOCK_OFF] = FIO_RANDSEED * td->thread_number;
init.c:1073: td->rand_seeds[i] = FIO_RANDSEED * td->thread_number
init.c:1235: td->rand_seeds[i] = seed * td->thread_number + i;
init.c:1565: td->thread_number, suf, o->per_job_logs);
init.c:1569: td->thread_number, suf, o->per_job_logs);
init.c:1573: td->thread_number, suf, o->per_job_logs);
init.c:1605: td->thread_number, suf, o->per_job_logs);
init.c:1637: td->thread_number, suf, o->per_job_logs);
init.c:1668: td->thread_number, suf, o->per_job_logs);
init.c:3004: if (!thread_number) {
libfio.c:160: thread_number = 0;
server.c:758: spdu.jobs = cpu_to_le32(thread_number);
server.c:801: spdu.jobs = cpu_to_le32(thread_number);
server.c:842: spdu.jobs = cpu_to_le32(thread_number);
server.c:943: tnumber = le32_to_cpu(pdu->thread_number);
server.c:947: if (!tnumber || tnumber > thread_number) {
server.c:1478: p.ts.thread_number = cpu_to_le32(ts->thread_number);
server.c:1958: .thread_number = cpu_to_le32(td->thread_number),
server.c:2029: .thread_number = cpu_to_le32(td->thread_number),
server.h:172: uint32_t thread_number;
server.h:192: uint32_t thread_number;
stat.c:1782: ts->thread_number = td->thread_number;
stat.c:1998: rt = malloc(thread_number * sizeof(unsigned long long));
stat.h:152: uint32_t thread_number;
stat.h:365:#define THREAD_RUNSTR_SZ __THREAD_RUNSTR_SZ(thread_number)
verify.c:1168: hdr->thread = td->thread_number;
verify.c:1797: fd = open_state_file(td->o.name, prefix, td->thread_number - 1, 0);$ grep -rHn thread_number HOWTO -C5
HOWTO-1240- offset is aligned to the minimum block size.
HOWTO-1241-
HOWTO-1242-.. option:: offset_increment=int
HOWTO-1243-
HOWTO-1244- If this is provided, then the real offset becomes `offset + offset_increment
HOWTO:1245: * thread_number`, where the thread number is a counter that starts at 0 and
HOWTO-1246- is incremented for each sub-job (i.e. when :option:`numjobs` option is
HOWTO-1247- specified). This option is useful if there are several jobs which are
HOWTO-1248- intended to operate on a file in parallel disjoint segments, with even
HOWTO-1249- spacing between the starting points.
HOWTO-1250-
$ grep -rHn thread_number init.c
init.c:480: if (thread_number >= max_jobs) {
init.c:486: td = &threads[thread_number++];
init.c:505: td->thread_number = thread_number;
init.c:536: memset(&threads[td->thread_number - 1], 0, sizeof(*td));
init.c:537: thread_number--;
init.c:1054: td->rand_seeds[FIO_RAND_BLOCK_OFF] = FIO_RANDSEED * td->thread_number;
init.c:1073: td->rand_seeds[i] = FIO_RANDSEED * td->thread_number
init.c:1235: td->rand_seeds[i] = seed * td->thread_number + i;
init.c:1565: td->thread_number, suf, o->per_job_logs);
init.c:1569: td->thread_number, suf, o->per_job_logs);
init.c:1573: td->thread_number, suf, o->per_job_logs);
init.c:1605: td->thread_number, suf, o->per_job_logs);
init.c:1637: td->thread_number, suf, o->per_job_logs);
init.c:1668: td->thread_number, suf, o->per_job_logs);
init.c:3004: if (!thread_number) {// init.c
/*
* Return a free job structure.
*/
static struct thread_data *get_new_job(bool global, struct thread_data *parent,
bool preserve_eo, const char *jobname)
{
struct thread_data *td;
if (global)
return &def_thread;
if (setup_thread_area()) {
log_err("error: failed to setup shm segment\n");
return NULL;
}
if (thread_number >= max_jobs) {
log_err("error: maximum number of jobs (%d) reached.\n",
max_jobs);
return NULL;
}
td = &threads[thread_number++];
*td = *parent;
INIT_FLIST_HEAD(&td->opt_list);
if (parent != &def_thread)
copy_opt_list(td, parent);
td->io_ops = NULL;
td->io_ops_init = 0;
if (!preserve_eo)
td->eo = NULL;
td->o.uid = td->o.gid = -1U;
dup_files(td, parent);
fio_options_mem_dupe(td);
profile_add_hooks(td);
td->thread_number = thread_number;
td->subjob_number = 0;
if (jobname)
td->o.name = strdup(jobname);
if (!parent->o.group_reporting || parent == &def_thread)
stat_number++;
return td;
}最终总结
整个随机序列的数值生成只依赖于创建这个Job的序号，以及整个测试序列的数量，这是实实在在的伪随机。之前的--debug=random中第一列是PID，第二列就是最终生成的序列序号；
先使用randwrite在文件系统中进行数据写入，如果文件系统为空，且写入期间没有其他文件被写入，那么虽然逻辑上是随机序列，但实际被文件系统分配到连续的物理扇区上（同时，写操作没有被块设备调度器进行合并，所以随机写的IO次数并未被变少）；
接着再进行randread进行IOPS测试时，由于伪随机的原因，和之前randwrite的逻辑随机序列是一样的，但经过文件系统到达块设备时，就是物理上连续的扇区，会被块设备调度器合并，变成顺序读，IO次数减少，所以这时测试的IOPS会偏大（至于为啥写没被合并，读却被合并，这跟文件系统和调度器算法中，读写合并等待时间以及队列长度等参数不同有关）；
在使用fio进行随机读测试之前，应该使用dd或者fio的顺序写初始化文件，使读写的序列顺序不同，即可规避这个问题；
至于随机写的测试，最好也要在测试前对文件系统进行一些列不规则的文件增删读写，使得文件系统的文件扇区分配不再连续，以便得出更靠谱的测试结果。
看完上述内容，你们掌握如何进行Fio随机读IOPS测试值可能偏大的原因分析的方法了吗？如果还想学到更多技能或想了解更多相关内容，欢迎关注天达云行业资讯频道，感谢各位的阅读！