X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=eta.c;h=238a0af13079c433daab4512ccabb45c9ce226cc;hp=a48cadae70684b43e901a2b9fc18bedf23d42504;hb=95820b6e6c92025df8d89c0bf39b174e53137c41;hpb=e6e41602c9725f811160f244b8ec5ecaa4912f45 diff --git a/eta.c b/eta.c index a48cadae..238a0af1 100644 --- a/eta.c +++ b/eta.c @@ -6,9 +6,8 @@ #include #include "fio.h" -#include "os.h" -static char run_str[MAX_JOBS + 1]; +static char run_str[REAL_MAX_JOBS + 1]; /* * Sets the status of the 'td' in the printed status map. @@ -18,47 +17,75 @@ static void check_str_update(struct thread_data *td) char c = run_str[td->thread_number - 1]; switch (td->runstate) { - case TD_REAPED: + case TD_REAPED: + if (td->error) + c = 'X'; + else if (td->sig) + c = 'K'; + else c = '_'; - break; - case TD_EXITED: - c = 'E'; - break; - case TD_RUNNING: - if (td_rw(td)) { - if (td_random(td)) - c = 'm'; - else - c = 'M'; - } else if (td_read(td)) { - if (td_random(td)) + break; + case TD_EXITED: + c = 'E'; + break; + case TD_RAMP: + c = '/'; + break; + case TD_RUNNING: + if (td_rw(td)) { + if (td_random(td)) { + if (td->o.rwmix[DDIR_READ] == 100) c = 'r'; - else - c = 'R'; - } else { - if (td_random(td)) + else if (td->o.rwmix[DDIR_WRITE] == 100) c = 'w'; else + c = 'm'; + } else { + if (td->o.rwmix[DDIR_READ] == 100) + c = 'R'; + else if (td->o.rwmix[DDIR_WRITE] == 100) c = 'W'; + else + c = 'M'; } - break; - case TD_VERIFYING: - c = 'V'; - break; - case TD_FSYNCING: - c = 'F'; - break; - case TD_CREATED: - c = 'C'; - break; - case TD_INITIALIZED: - c = 'I'; - break; - case TD_NOT_CREATED: - c = 'P'; - break; - default: - log_err("state %d\n", td->runstate); + } else if (td_read(td)) { + if (td_random(td)) + c = 'r'; + else + c = 'R'; + } else if (td_write(td)) { + if (td_random(td)) + c = 'w'; + else + c = 'W'; + } else { + if (td_random(td)) + c = 'd'; + else + c = 'D'; + } + break; + case TD_PRE_READING: + c = 'p'; + break; + case TD_VERIFYING: + c = 'V'; + break; + case TD_FSYNCING: + c = 'F'; + break; + case TD_CREATED: + c = 'C'; + break; + case TD_INITIALIZED: + case TD_SETTING_UP: + c = 'I'; + break; + case TD_NOT_CREATED: + c = 'P'; + break; + default: + log_err("state %d\n", td->runstate); } run_str[td->thread_number - 1] = c; @@ -67,13 +94,11 @@ static void check_str_update(struct thread_data *td) /* * Convert seconds to a printable string. */ -static void eta_to_str(char *str, int eta_sec) +void eta_to_str(char *str, unsigned long eta_sec) { unsigned int d, h, m, s; int disp_hour = 0; - d = h = m = s = 0; - s = eta_sec % 60; eta_sec /= 60; m = eta_sec % 60; @@ -97,21 +122,37 @@ static void eta_to_str(char *str, int eta_sec) /* * Best effort calculation of the estimated pending runtime of a job. */ -static int thread_eta(struct thread_data *td, unsigned long elapsed) +static int thread_eta(struct thread_data *td) { unsigned long long bytes_total, bytes_done; unsigned long eta_sec = 0; + unsigned long elapsed; + + elapsed = (mtime_since_now(&td->epoch) + 999) / 1000; bytes_total = td->total_io_size; + if (td->o.fill_device && td->o.size == -1ULL) { + if (!td->fill_device_size || td->fill_device_size == -1ULL) + return 0; + + bytes_total = td->fill_device_size; + } + /* - * if writing, bytes_total will be twice the size. If mixing, - * assume a 50/50 split and thus bytes_total will be 50% larger. + * if writing and verifying afterwards, bytes_total will be twice the + * size. In a mixed workload, verify phase will be the size of the + * first stage writes. */ - if (td->o.verify) { - if (td_rw(td)) - bytes_total = bytes_total * 3 / 2; - else + if (td->o.do_verify && td->o.verify && td_write(td)) { + if (td_rw(td)) { + unsigned int perc = 50; + + if (td->o.rwmix[DDIR_WRITE]) + perc = td->o.rwmix[DDIR_WRITE]; + + bytes_total += (bytes_total * perc) / 100; + } else bytes_total <<= 1; } @@ -119,30 +160,52 @@ static int thread_eta(struct thread_data *td, unsigned long elapsed) bytes_total /= (td->o.zone_skip / td->o.zone_size); if (td->runstate == TD_RUNNING || td->runstate == TD_VERIFYING) { - double perc; + double perc, perc_t; - bytes_done = td->io_bytes[DDIR_READ] + td->io_bytes[DDIR_WRITE]; + bytes_done = ddir_rw_sum(td->io_bytes); perc = (double) bytes_done / (double) bytes_total; if (perc > 1.0) perc = 1.0; + if (td->o.time_based) { + perc_t = (double) elapsed / (double) td->o.timeout; + if (perc_t < perc) + perc = perc_t; + } + eta_sec = (unsigned long) (elapsed * (1.0 / perc)) - elapsed; - if (td->o.timeout && eta_sec > (td->o.timeout - elapsed)) - eta_sec = td->o.timeout - elapsed; + if (td->o.timeout && + eta_sec > (td->o.timeout + done_secs - elapsed)) + eta_sec = td->o.timeout + done_secs - elapsed; } else if (td->runstate == TD_NOT_CREATED || td->runstate == TD_CREATED - || td->runstate == TD_INITIALIZED) { + || td->runstate == TD_INITIALIZED + || td->runstate == TD_RAMP + || td->runstate == TD_PRE_READING) { int t_eta = 0, r_eta = 0; + unsigned long long rate_bytes; /* * We can only guess - assume it'll run the full timeout * if given, otherwise assume it'll run at the specified rate. */ - if (td->o.timeout) - t_eta = td->o.timeout + td->o.start_delay - elapsed; - if (td->o.rate) { - r_eta = (bytes_total / 1024) / td->o.rate; - r_eta += td->o.start_delay - elapsed; + if (td->o.timeout) { + t_eta = td->o.timeout + td->o.start_delay + + td->o.ramp_time; + + if (in_ramp_time(td)) { + unsigned long ramp_left; + + ramp_left = mtime_since_now(&td->epoch); + ramp_left = (ramp_left + 999) / 1000; + if (ramp_left <= t_eta) + t_eta -= ramp_left; + } + } + rate_bytes = ddir_rw_sum(td->o.rate); + if (rate_bytes) { + r_eta = (bytes_total / 1024) / rate_bytes; + r_eta += td->o.start_delay; } if (r_eta && t_eta) @@ -163,131 +226,290 @@ static int thread_eta(struct thread_data *td, unsigned long elapsed) return eta_sec; } -static void calc_rate(unsigned long mtime, unsigned long long *io_bytes, +static void calc_rate(int unified_rw_rep, unsigned long mtime, + unsigned long long *io_bytes, unsigned long long *prev_io_bytes, unsigned int *rate) { - rate[0] = (io_bytes[0] - prev_io_bytes[0]) / mtime; - rate[1] = (io_bytes[1] - prev_io_bytes[1]) / mtime; - prev_io_bytes[0] = io_bytes[0]; - prev_io_bytes[1] = io_bytes[1]; + int i; + + for (i = 0; i < DDIR_RWDIR_CNT; i++) { + unsigned long long diff; + + diff = io_bytes[i] - prev_io_bytes[i]; + if (unified_rw_rep) { + rate[i] = 0; + rate[0] += ((1000 * diff) / mtime) / 1024; + } else + rate[i] = ((1000 * diff) / mtime) / 1024; + + prev_io_bytes[i] = io_bytes[i]; + } +} + +static void calc_iops(int unified_rw_rep, unsigned long mtime, + unsigned long long *io_iops, + unsigned long long *prev_io_iops, unsigned int *iops) +{ + int i; + + for (i = 0; i < DDIR_RWDIR_CNT; i++) { + unsigned long long diff; + + diff = io_iops[i] - prev_io_iops[i]; + if (unified_rw_rep) { + iops[i] = 0; + iops[0] += (diff * 1000) / mtime; + } else + iops[i] = (diff * 1000) / mtime; + + prev_io_iops[i] = io_iops[i]; + } } - + /* * Print status of the jobs we know about. This includes rate estimates, * ETA, thread state, etc. */ -void print_thread_status(void) +int calc_thread_status(struct jobs_eta *je, int force) { - unsigned long elapsed = mtime_since_genesis() / 1000; - int i, nr_running, nr_pending, t_rate, m_rate, *eta_secs, eta_sec; - int t_iops, m_iops, files_open; struct thread_data *td; - char eta_str[128]; - double perc = 0.0; - unsigned long long io_bytes[2]; - unsigned long rate_time, disp_time, bw_avg_time; + int i, unified_rw_rep; + unsigned long rate_time, disp_time, bw_avg_time, *eta_secs; + unsigned long long io_bytes[DDIR_RWDIR_CNT]; + unsigned long long io_iops[DDIR_RWDIR_CNT]; struct timeval now; - static unsigned long long rate_io_bytes[2]; - static unsigned long long disp_io_bytes[2]; + static unsigned long long rate_io_bytes[DDIR_RWDIR_CNT]; + static unsigned long long disp_io_bytes[DDIR_RWDIR_CNT]; + static unsigned long long disp_io_iops[DDIR_RWDIR_CNT]; static struct timeval rate_prev_time, disp_prev_time; - static unsigned int rate[2]; - if (temp_stall_ts || terse_output) - return; + if (!force) { + if (output_format != FIO_OUTPUT_NORMAL) + return 0; + if (temp_stall_ts || eta_print == FIO_ETA_NEVER) + return 0; - if (!rate_io_bytes[0] && !rate_io_bytes[1]) + if (!isatty(STDOUT_FILENO) && (eta_print != FIO_ETA_ALWAYS)) + return 0; + } + + if (!ddir_rw_sum(rate_io_bytes)) fill_start_time(&rate_prev_time); - if (!disp_io_bytes[0] && !disp_io_bytes[1]) + if (!ddir_rw_sum(disp_io_bytes)) fill_start_time(&disp_prev_time); - eta_secs = malloc(thread_number * sizeof(int)); - memset(eta_secs, 0, thread_number * sizeof(int)); + eta_secs = malloc(thread_number * sizeof(unsigned long)); + memset(eta_secs, 0, thread_number * sizeof(unsigned long)); + + je->elapsed_sec = (mtime_since_genesis() + 999) / 1000; - io_bytes[0] = io_bytes[1] = 0; - nr_pending = nr_running = t_rate = m_rate = t_iops = m_iops = 0; + io_bytes[DDIR_READ] = io_bytes[DDIR_WRITE] = io_bytes[DDIR_TRIM] = 0; + io_iops[DDIR_READ] = io_iops[DDIR_WRITE] = io_iops[DDIR_TRIM] = 0; bw_avg_time = ULONG_MAX; - files_open = 0; + unified_rw_rep = 0; for_each_td(td, i) { + unified_rw_rep += td->o.unified_rw_rep; + if (is_power_of_2(td->o.kb_base)) + je->is_pow2 = 1; if (td->o.bw_avg_time < bw_avg_time) bw_avg_time = td->o.bw_avg_time; if (td->runstate == TD_RUNNING || td->runstate == TD_VERIFYING - || td->runstate == TD_FSYNCING) { - nr_running++; - t_rate += td->o.rate; - m_rate += td->o.ratemin; - t_iops += td->o.rate_iops; - m_iops += td->o.rate_iops_min; - files_open += td->nr_open_files; - } else if (td->runstate < TD_RUNNING) - nr_pending++; - - if (elapsed >= 3) - eta_secs[i] = thread_eta(td, elapsed); + || td->runstate == TD_FSYNCING + || td->runstate == TD_PRE_READING) { + je->nr_running++; + if (td_read(td)) { + je->t_rate[0] += td->o.rate[DDIR_READ]; + je->t_iops[0] += td->o.rate_iops[DDIR_READ]; + je->m_rate[0] += td->o.ratemin[DDIR_READ]; + je->m_iops[0] += td->o.rate_iops_min[DDIR_READ]; + } + if (td_write(td)) { + je->t_rate[1] += td->o.rate[DDIR_WRITE]; + je->t_iops[1] += td->o.rate_iops[DDIR_WRITE]; + je->m_rate[1] += td->o.ratemin[DDIR_WRITE]; + je->m_iops[1] += td->o.rate_iops_min[DDIR_WRITE]; + } + if (td_trim(td)) { + je->t_rate[2] += td->o.rate[DDIR_TRIM]; + je->t_iops[2] += td->o.rate_iops[DDIR_TRIM]; + je->m_rate[2] += td->o.ratemin[DDIR_TRIM]; + je->m_iops[2] += td->o.rate_iops_min[DDIR_TRIM]; + } + + je->files_open += td->nr_open_files; + } else if (td->runstate == TD_RAMP) { + je->nr_running++; + je->nr_ramp++; + } else if (td->runstate == TD_SETTING_UP) + je->nr_running++; + else if (td->runstate < TD_RUNNING) + je->nr_pending++; + + if (je->elapsed_sec >= 3) + eta_secs[i] = thread_eta(td); else eta_secs[i] = INT_MAX; check_str_update(td); - io_bytes[0] += td->io_bytes[0]; - io_bytes[1] += td->io_bytes[1]; + + if (td->runstate > TD_RAMP) { + int ddir; + + for (ddir = DDIR_READ; ddir < DDIR_RWDIR_CNT; ddir++) { + if (unified_rw_rep) { + io_bytes[0] += td->io_bytes[ddir]; + io_iops[0] += td->io_blocks[ddir]; + } else { + io_bytes[ddir] += td->io_bytes[ddir]; + io_iops[ddir] += td->io_blocks[ddir]; + } + } + } } if (exitall_on_terminate) - eta_sec = INT_MAX; + je->eta_sec = INT_MAX; else - eta_sec = 0; + je->eta_sec = 0; for_each_td(td, i) { if (exitall_on_terminate) { - if (eta_secs[i] < eta_sec) - eta_sec = eta_secs[i]; + if (eta_secs[i] < je->eta_sec) + je->eta_sec = eta_secs[i]; } else { - if (eta_secs[i] > eta_sec) - eta_sec = eta_secs[i]; + if (eta_secs[i] > je->eta_sec) + je->eta_sec = eta_secs[i]; } } free(eta_secs); - if (eta_sec != INT_MAX && elapsed) { - perc = (double) elapsed / (double) (elapsed + eta_sec); - eta_to_str(eta_str, eta_sec); - } - fio_gettime(&now, NULL); rate_time = mtime_since(&rate_prev_time, &now); - if (write_bw_log && rate_time> bw_avg_time) { - calc_rate(rate_time, io_bytes, rate_io_bytes, rate); + if (write_bw_log && rate_time > bw_avg_time && !in_ramp_time(td)) { + calc_rate(unified_rw_rep, rate_time, io_bytes, rate_io_bytes, + je->rate); memcpy(&rate_prev_time, &now, sizeof(now)); - add_agg_sample(rate[DDIR_READ], DDIR_READ); - add_agg_sample(rate[DDIR_WRITE], DDIR_WRITE); + add_agg_sample(je->rate[DDIR_READ], DDIR_READ, 0); + add_agg_sample(je->rate[DDIR_WRITE], DDIR_WRITE, 0); + add_agg_sample(je->rate[DDIR_TRIM], DDIR_TRIM, 0); } disp_time = mtime_since(&disp_prev_time, &now); - if (disp_time < 1000) - return; - calc_rate(disp_time, io_bytes, disp_io_bytes, rate); + /* + * Allow a little slack, the target is to print it every 1000 msecs + */ + if (!force && disp_time < 900) + return 0; + + calc_rate(unified_rw_rep, disp_time, io_bytes, disp_io_bytes, je->rate); + calc_iops(unified_rw_rep, disp_time, io_iops, disp_io_iops, je->iops); + memcpy(&disp_prev_time, &now, sizeof(now)); - if (!nr_running && !nr_pending) - return; + if (!force && !je->nr_running && !je->nr_pending) + return 0; + + je->nr_threads = thread_number; + memcpy(je->run_str, run_str, thread_number * sizeof(char)); + + return 1; +} - printf("Jobs: %d (f=%d)", nr_running, files_open); - if (m_rate || t_rate) - printf(", CR=%d/%d KiB/s", t_rate, m_rate); - else if (m_iops || t_iops) - printf(", CR=%d/%d IOPS", t_iops, m_iops); - if (eta_sec != INT_MAX && nr_running) { - perc *= 100.0; - printf(": [%s] [%3.1f%% done] [%6u/%6u kb/s] [eta %s]", run_str, perc, rate[0], rate[1], eta_str); +void display_thread_status(struct jobs_eta *je) +{ + static int linelen_last; + static int eta_good; + char output[REAL_MAX_JOBS + 512], *p = output; + char eta_str[128]; + double perc = 0.0; + + if (je->eta_sec != INT_MAX && je->elapsed_sec) { + perc = (double) je->elapsed_sec / (double) (je->elapsed_sec + je->eta_sec); + eta_to_str(eta_str, je->eta_sec); + } + + p += sprintf(p, "Jobs: %d (f=%d)", je->nr_running, je->files_open); + if (je->m_rate[0] || je->m_rate[1] || je->t_rate[0] || je->t_rate[1]) { + char *tr, *mr; + + mr = num2str(je->m_rate[0] + je->m_rate[1], 4, 0, je->is_pow2); + tr = num2str(je->t_rate[0] + je->t_rate[1], 4, 0, je->is_pow2); + p += sprintf(p, ", CR=%s/%s KB/s", tr, mr); + free(tr); + free(mr); + } else if (je->m_iops[0] || je->m_iops[1] || je->t_iops[0] || je->t_iops[1]) { + p += sprintf(p, ", CR=%d/%d IOPS", + je->t_iops[0] + je->t_iops[1], + je->m_iops[0] + je->m_iops[1]); + } + if (je->eta_sec != INT_MAX && je->nr_running) { + char perc_str[32]; + char *iops_str[DDIR_RWDIR_CNT]; + char *rate_str[DDIR_RWDIR_CNT]; + size_t left; + int l; + int ddir; + + if ((!je->eta_sec && !eta_good) || je->nr_ramp == je->nr_running) + strcpy(perc_str, "-.-% done"); + else { + eta_good = 1; + perc *= 100.0; + sprintf(perc_str, "%3.1f%% done", perc); + } + + for (ddir = DDIR_READ; ddir < DDIR_RWDIR_CNT; ddir++) { + rate_str[ddir] = num2str(je->rate[ddir], 5, + 1024, je->is_pow2); + iops_str[ddir] = num2str(je->iops[ddir], 4, 1, 0); + } + + left = sizeof(output) - (p - output) - 1; + + l = snprintf(p, left, ": [%s] [%s] [%s/%s/%s /s] [%s/%s/%s iops] [eta %s]", + je->run_str, perc_str, rate_str[DDIR_READ], + rate_str[DDIR_WRITE], rate_str[DDIR_TRIM], + iops_str[DDIR_READ], iops_str[DDIR_WRITE], + iops_str[DDIR_TRIM], eta_str); + p += l; + if (l >= 0 && l < linelen_last) + p += sprintf(p, "%*s", linelen_last - l, ""); + linelen_last = l; + + for (ddir = DDIR_READ; ddir < DDIR_RWDIR_CNT; ddir++) { + free(rate_str[ddir]); + free(iops_str[ddir]); + } } - printf("\r"); + p += sprintf(p, "\r"); + + printf("%s", output); fflush(stdout); } -void print_status_init(int thread_number) +void print_thread_status(void) +{ + struct jobs_eta *je; + size_t size; + + if (!thread_number) + return; + + size = sizeof(*je) + thread_number * sizeof(char) + 1; + je = malloc(size); + memset(je, 0, size); + + if (calc_thread_status(je, 0)) + display_thread_status(je); + + free(je); +} + +void print_status_init(int thr_number) { - run_str[thread_number] = 'P'; + run_str[thr_number] = 'P'; }