X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fobdfilter%2Flproc_obdfilter.c;h=e25568c90b2e6e7222f4cc884e6f53c8fb3b6d93;hp=ef632a959809b2a2aff8d6b923b36e18a9138931;hb=1a24137e8f26eaae9a2dac39a1e8a8a0bed46b6b;hpb=fbb7ead129258897f5a5d5c9ce28d31fbbe5bca2 diff --git a/lustre/obdfilter/lproc_obdfilter.c b/lustre/obdfilter/lproc_obdfilter.c index ef632a9..e25568c 100644 --- a/lustre/obdfilter/lproc_obdfilter.c +++ b/lustre/obdfilter/lproc_obdfilter.c @@ -1,44 +1,53 @@ /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- * vim:expandtab:shiftwidth=8:tabstop=8: * - * Copyright (C) 2002, 2003 Cluster File Systems, Inc. + * GPL HEADER START * - * This file is part of Lustre, http://www.lustre.org. + * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * - * Lustre is free software; you can redistribute it and/or - * modify it under the terms of version 2 of the GNU General Public - * License as published by the Free Software Foundation. + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 only, + * as published by the Free Software Foundation. * - * Lustre is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License version 2 for more details (a copy is included + * in the LICENSE file that accompanied this code). * - * You should have received a copy of the GNU General Public License - * along with Lustre; if not, write to the Free Software - * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * You should have received a copy of the GNU General Public License + * version 2 along with this program; If not, see + * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf * + * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, + * CA 95054 USA or visit www.sun.com if you need additional information or + * have any questions. + * + * GPL HEADER END + */ +/* + * Copyright 2008 Sun Microsystems, Inc. All rights reserved + * Use is subject to license terms. + */ +/* + * This file is part of Lustre, http://www.lustre.org/ + * Lustre is a trademark of Sun Microsystems, Inc. */ #define DEBUG_SUBSYSTEM S_CLASS #include -#include -#include +#include +#include #include #include #include "filter_internal.h" -#ifndef LPROCFS -static struct lprocfs_vars lprocfs_obd_vars[] = { {0} }; -static struct lprocfs_vars lprocfs_module_vars[] = { {0} }; -#else - +#ifdef LPROCFS static int lprocfs_filter_rd_groups(char *page, char **start, off_t off, int count, int *eof, void *data) { struct obd_device *obd = (struct obd_device *)data; - *eof = 1; return snprintf(page, count, "%u\n", obd->u.filter.fo_group_count); } @@ -95,7 +104,7 @@ static int lprocfs_filter_rd_last_id(char *page, char **start, off_t off, if (obd == NULL) return 0; - for (i = 1; i < filter->fo_group_count; i++) { + for (i = FILTER_GROUP_MDS0; i < filter->fo_group_count; i++) { rc = snprintf(page, count, LPU64"\n",filter_last_id(filter, i)); if (rc < 0) { retval = rc; @@ -134,7 +143,106 @@ int lprocfs_filter_wr_readcache(struct file *file, const char *buffer, return count; } -static struct lprocfs_vars lprocfs_obd_vars[] = { + +int lprocfs_filter_rd_fmd_max_num(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct obd_device *obd = data; + int rc; + + rc = snprintf(page, count, "%u\n", obd->u.filter.fo_fmd_max_num); + return rc; +} + +int lprocfs_filter_wr_fmd_max_num(struct file *file, const char *buffer, + unsigned long count, void *data) +{ + struct obd_device *obd = data; + int val; + int rc; + + rc = lprocfs_write_helper(buffer, count, &val); + if (rc) + return rc; + + if (val > 65536 || val < 1) + return -EINVAL; + + obd->u.filter.fo_fmd_max_num = val; + return count; +} + +int lprocfs_filter_rd_fmd_max_age(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct obd_device *obd = data; + int rc; + + rc = snprintf(page, count, "%u\n", obd->u.filter.fo_fmd_max_age / HZ); + return rc; +} + +int lprocfs_filter_wr_fmd_max_age(struct file *file, const char *buffer, + unsigned long count, void *data) +{ + struct obd_device *obd = data; + int val; + int rc; + + rc = lprocfs_write_helper(buffer, count, &val); + if (rc) + return rc; + + if (val > 65536 || val < 1) + return -EINVAL; + + obd->u.filter.fo_fmd_max_age = val * HZ; + return count; +} + +static int lprocfs_filter_rd_capa(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + struct obd_device *obd = data; + int rc; + + rc = snprintf(page, count, "capability on: %s\n", + obd->u.filter.fo_fl_oss_capa ? "oss" : ""); + return rc; +} + +static int lprocfs_filter_wr_capa(struct file *file, const char *buffer, + unsigned long count, void *data) +{ + struct obd_device *obd = data; + int val, rc; + + rc = lprocfs_write_helper(buffer, count, &val); + if (rc) + return rc; + + if (val & ~0x1) { + CERROR("invalid capability mode, only 0/1 are accepted.\n" + " 1: enable oss fid capability\n" + " 0: disable oss fid capability\n"); + return -EINVAL; + } + + obd->u.filter.fo_fl_oss_capa = val; + LCONSOLE_INFO("OSS %s %s fid capability.\n", obd->obd_name, + val ? "enabled" : "disabled"); + return count; +} + +static int lprocfs_filter_rd_capa_count(char *page, char **start, off_t off, + int count, int *eof, void *data) +{ + return snprintf(page, count, "%d %d\n", + capa_count[CAPA_SITE_CLIENT], + capa_count[CAPA_SITE_SERVER]); +} + +static struct lprocfs_vars lprocfs_filter_obd_vars[] = { { "uuid", lprocfs_rd_uuid, 0, 0 }, { "blocksize", lprocfs_rd_blksize, 0, 0 }, { "kbytestotal", lprocfs_rd_kbytestotal, 0, 0 }, @@ -149,22 +257,43 @@ static struct lprocfs_vars lprocfs_obd_vars[] = { { "tot_dirty", lprocfs_filter_rd_tot_dirty, 0, 0 }, { "tot_pending", lprocfs_filter_rd_tot_pending, 0, 0 }, { "tot_granted", lprocfs_filter_rd_tot_granted, 0, 0 }, + { "hash_stats", lprocfs_obd_rd_hash, 0, 0 }, + { "recovery_status", lprocfs_obd_rd_recovery_status, 0, 0 }, + { "recovery_maxtime", lprocfs_obd_rd_recovery_maxtime, + lprocfs_obd_wr_recovery_maxtime, 0}, + { "evict_client", 0, lprocfs_wr_evict_client, 0, + &lprocfs_evict_client_fops}, { "num_exports", lprocfs_rd_num_exports, 0, 0 }, { "readcache_max_filesize", lprocfs_filter_rd_readcache, lprocfs_filter_wr_readcache, 0 }, - { "recovery_status", lprocfs_obd_rd_recovery_status, 0, 0 }, +#ifdef HAVE_QUOTA_SUPPORT + { "quota_bunit_sz", lprocfs_rd_bunit, lprocfs_wr_bunit, 0}, + { "quota_btune_sz", lprocfs_rd_btune, lprocfs_wr_btune, 0}, + { "quota_iunit_sz", lprocfs_rd_iunit, lprocfs_wr_iunit, 0}, + { "quota_itune_sz", lprocfs_rd_itune, lprocfs_wr_itune, 0}, + { "quota_type", lprocfs_rd_type, lprocfs_wr_type, 0}, +#endif + { "client_cache_count", lprocfs_filter_rd_fmd_max_num, + lprocfs_filter_wr_fmd_max_num, 0 }, + { "client_cache_seconds", lprocfs_filter_rd_fmd_max_age, + lprocfs_filter_wr_fmd_max_age, 0 }, + { "capa", lprocfs_filter_rd_capa, + lprocfs_filter_wr_capa, 0 }, + { "capa_count", lprocfs_filter_rd_capa_count, 0, 0 }, { 0 } }; -static struct lprocfs_vars lprocfs_module_vars[] = { +static struct lprocfs_vars lprocfs_filter_module_vars[] = { { "num_refs", lprocfs_rd_numrefs, 0, 0 }, { 0 } }; -void filter_tally_write(struct filter_obd *filter, struct page **pages, - int nr_pages, unsigned long *blocks, int blocks_per_page) +void filter_tally(struct obd_export *exp, struct page **pages, int nr_pages, + unsigned long *blocks, int blocks_per_page, int wr) { + struct filter_obd *filter = &exp->exp_obd->u.filter; + struct filter_export_data *fed = &exp->exp_filter_data; struct page *last_page = NULL; unsigned long *last_block = NULL; unsigned long discont_pages = 0; @@ -174,7 +303,13 @@ void filter_tally_write(struct filter_obd *filter, struct page **pages, if (nr_pages == 0) return; - lprocfs_oh_tally_log2(&filter->fo_w_pages, nr_pages); + lprocfs_oh_tally_log2(&filter->fo_filter_stats.hist[BRW_R_PAGES + wr], + nr_pages); + lprocfs_oh_tally_log2(&fed->fed_brw_stats.hist[BRW_R_PAGES + wr], + nr_pages); + if (exp->exp_nid_stats && exp->exp_nid_stats->nid_brw_stats) + lprocfs_oh_tally_log2(&exp->exp_nid_stats->nid_brw_stats->hist[BRW_W_PAGES + wr], + nr_pages); while (nr_pages-- > 0) { if (last_page && (*pages)->index != (last_page->index + 1)) @@ -188,167 +323,114 @@ void filter_tally_write(struct filter_obd *filter, struct page **pages, } } - lprocfs_oh_tally(&filter->fo_w_discont_pages, discont_pages); - lprocfs_oh_tally(&filter->fo_w_discont_blocks, discont_blocks); -} - -void filter_tally_read(struct filter_obd *filter, struct niobuf_local *lnb, - int niocount) -{ - struct niobuf_local *end; - struct page *last_page = NULL; - unsigned long discont_pages = 0; - unsigned long discont_blocks = 0; - - if (niocount == 0) - return; - - for (end = lnb + niocount; lnb < end && lnb->page; lnb++) { - struct page *page = lnb->page; - if (last_page) { - if (page->index != (last_page->index + 1)) - discont_pages++; - /* XXX not so smart for now */ -#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0)) - if ((page->buffers && last_page->buffers) && - (page->buffers->b_blocknr != - (last_page->buffers->b_blocknr + 1))) - discont_blocks++; -#else -#warning "port on 2.6 -bzzz" -#endif - } - last_page = page; + lprocfs_oh_tally(&filter->fo_filter_stats.hist[BRW_R_DISCONT_PAGES +wr], + discont_pages); + lprocfs_oh_tally(&fed->fed_brw_stats.hist[BRW_R_DISCONT_PAGES + wr], + discont_pages); + lprocfs_oh_tally(&filter->fo_filter_stats.hist[BRW_R_DISCONT_BLOCKS+wr], + discont_blocks); + lprocfs_oh_tally(&fed->fed_brw_stats.hist[BRW_R_DISCONT_BLOCKS + wr], + discont_blocks); + + if (exp->exp_nid_stats && exp->exp_nid_stats->nid_brw_stats) { + lprocfs_oh_tally_log2(&exp->exp_nid_stats->nid_brw_stats->hist[BRW_W_DISCONT_PAGES + wr], + discont_pages); + lprocfs_oh_tally_log2(&exp->exp_nid_stats->nid_brw_stats->hist[BRW_W_DISCONT_BLOCKS + wr], + discont_blocks); } - - lprocfs_oh_tally_log2(&filter->fo_r_pages, niocount); - lprocfs_oh_tally(&filter->fo_r_discont_pages, discont_pages); - lprocfs_oh_tally(&filter->fo_r_discont_blocks, discont_blocks); } #define pct(a,b) (b ? a * 100 / b : 0) -static int filter_brw_stats_seq_show(struct seq_file *seq, void *v) +static void display_brw_stats(struct seq_file *seq, char *name, char *units, + struct obd_histogram *read, struct obd_histogram *write, int log2) { - struct timeval now; - struct obd_device *dev = seq->private; - struct filter_obd *filter = &dev->u.filter; - unsigned long read_tot = 0, write_tot = 0, read_cum, write_cum; + unsigned long read_tot, write_tot, r, w, read_cum = 0, write_cum = 0; int i; - do_gettimeofday(&now); - - /* this sampling races with updates */ - - seq_printf(seq, "snapshot_time: %lu.%lu (secs.usecs)\n", - now.tv_sec, now.tv_usec); - - seq_printf(seq, "\n\t\t\tread\t\t\twrite\n"); - seq_printf(seq, "pages per brw brws %% cum %% |"); - seq_printf(seq, " rpcs %% cum %%\n"); + seq_printf(seq, "\n%26s read | write\n", " "); + seq_printf(seq, "%-22s %-5s %% cum %% | %-5s %% cum %%\n", + name, units, units); - read_tot = lprocfs_oh_sum(&filter->fo_r_pages); - write_tot = lprocfs_oh_sum(&filter->fo_w_pages); - - read_cum = 0; - write_cum = 0; + read_tot = lprocfs_oh_sum(read); + write_tot = lprocfs_oh_sum(write); for (i = 0; i < OBD_HIST_MAX; i++) { - unsigned long r = filter->fo_r_pages.oh_buckets[i]; - unsigned long w = filter->fo_w_pages.oh_buckets[i]; + r = read->oh_buckets[i]; + w = write->oh_buckets[i]; read_cum += r; write_cum += w; - seq_printf(seq, "%d:\t\t%10lu %3lu %3lu | %10lu %3lu %3lu\n", - 1 << i, r, pct(r, read_tot), - pct(read_cum, read_tot), w, - pct(w, write_tot), - pct(write_cum, write_tot)); + if (read_cum == 0 && write_cum == 0) + continue; + + if (!log2) + seq_printf(seq, "%u", i); + else if (i < 10) + seq_printf(seq, "%u", 1<fo_r_discont_pages); - write_tot = lprocfs_oh_sum(&filter->fo_w_discont_pages); + /* this sampling races with updates */ + do_gettimeofday(&now); + seq_printf(seq, "snapshot_time: %lu.%lu (secs.usecs)\n", + now.tv_sec, now.tv_usec); - read_cum = 0; - write_cum = 0; + display_brw_stats(seq, "pages per bulk r/w", "rpcs", + &brw_stats->hist[BRW_R_PAGES], + &brw_stats->hist[BRW_W_PAGES], 1); - for (i = 0; i < OBD_HIST_MAX; i++) { - unsigned long r = filter->fo_r_discont_pages.oh_buckets[i]; - unsigned long w = filter->fo_w_discont_pages.oh_buckets[i]; - read_cum += r; - write_cum += w; - seq_printf(seq, "%d:\t\t%10lu %3lu %3lu | %10lu %3lu %3lu\n", - i, r, pct(r, read_tot), - pct(read_cum, read_tot), w, - pct(w, write_tot), - pct(write_cum, write_tot)); - if (read_cum == read_tot && write_cum == write_tot) - break; - } + display_brw_stats(seq, "discontiguous pages", "rpcs", + &brw_stats->hist[BRW_R_DISCONT_PAGES], + &brw_stats->hist[BRW_W_DISCONT_PAGES], 0); - seq_printf(seq, "\n\t\t\tread\t\t\twrite\n"); - seq_printf(seq, "discont blocks rpcs %% cum %% |"); - seq_printf(seq, " rpcs %% cum %%\n"); + display_brw_stats(seq, "discontiguous blocks", "rpcs", + &brw_stats->hist[BRW_R_DISCONT_BLOCKS], + &brw_stats->hist[BRW_W_DISCONT_BLOCKS], 0); - read_tot = lprocfs_oh_sum(&filter->fo_r_discont_blocks); - write_tot = lprocfs_oh_sum(&filter->fo_w_discont_blocks); + display_brw_stats(seq, "disk fragmented I/Os", "ios", + &brw_stats->hist[BRW_R_DIO_FRAGS], + &brw_stats->hist[BRW_W_DIO_FRAGS], 0); - read_cum = 0; - write_cum = 0; - for (i = 0; i < OBD_HIST_MAX; i++) { - unsigned long r = filter->fo_r_discont_blocks.oh_buckets[i]; - unsigned long w = filter->fo_w_discont_blocks.oh_buckets[i]; - read_cum += r; - write_cum += w; - seq_printf(seq, "%d:\t\t%10lu %3lu %3lu | %10lu %3lu %3lu\n", - i, r, pct(r, read_tot), - pct(read_cum, read_tot), w, - pct(w, write_tot), - pct(write_cum, write_tot)); - if (read_cum == read_tot && write_cum == write_tot) - break; + display_brw_stats(seq, "disk I/Os in flight", "ios", + &brw_stats->hist[BRW_R_RPC_HIST], + &brw_stats->hist[BRW_W_RPC_HIST], 0); + + { + char title[24]; + sprintf(title, "I/O time (1/%ds)", HZ); + display_brw_stats(seq, title, "ios", + &brw_stats->hist[BRW_R_IO_TIME], + &brw_stats->hist[BRW_W_IO_TIME], 1); } - return 0; + display_brw_stats(seq, "disk I/O size", "ios", + &brw_stats->hist[BRW_R_DISK_IOSIZE], + &brw_stats->hist[BRW_W_DISK_IOSIZE], 1); } + #undef pct -static void *filter_brw_stats_seq_start(struct seq_file *p, loff_t *pos) -{ - if (*pos == 0) - return (void *)1; - return NULL; -} -static void *filter_brw_stats_seq_next(struct seq_file *p, void *v, loff_t *pos) -{ - ++*pos; - return NULL; -} -static void filter_brw_stats_seq_stop(struct seq_file *p, void *v) +static int filter_brw_stats_seq_show(struct seq_file *seq, void *v) { -} -struct seq_operations filter_brw_stats_seq_sops = { - .start = filter_brw_stats_seq_start, - .stop = filter_brw_stats_seq_stop, - .next = filter_brw_stats_seq_next, - .show = filter_brw_stats_seq_show, -}; + struct obd_device *dev = seq->private; + struct filter_obd *filter = &dev->u.filter; -static int filter_brw_stats_seq_open(struct inode *inode, struct file *file) -{ - struct proc_dir_entry *dp = PDE(inode); - struct seq_file *seq; - int rc; + brw_stats_show(seq, &filter->fo_filter_stats); - rc = seq_open(file, &filter_brw_stats_seq_sops); - if (rc) - return rc; - seq = file->private_data; - seq->private = dp->data; return 0; } @@ -358,25 +440,15 @@ static ssize_t filter_brw_stats_seq_write(struct file *file, const char *buf, struct seq_file *seq = file->private_data; struct obd_device *dev = seq->private; struct filter_obd *filter = &dev->u.filter; + int i; - lprocfs_oh_clear(&filter->fo_r_pages); - lprocfs_oh_clear(&filter->fo_w_pages); - lprocfs_oh_clear(&filter->fo_r_discont_pages); - lprocfs_oh_clear(&filter->fo_w_discont_pages); - lprocfs_oh_clear(&filter->fo_r_discont_blocks); - lprocfs_oh_clear(&filter->fo_w_discont_blocks); + for (i = 0; i < BRW_LAST; i++) + lprocfs_oh_clear(&filter->fo_filter_stats.hist[i]); return len; } -struct file_operations filter_brw_stats_fops = { - .owner = THIS_MODULE, - .open = filter_brw_stats_seq_open, - .read = seq_read, - .write = filter_brw_stats_seq_write, - .llseek = seq_lseek, - .release = seq_release, -}; +LPROC_SEQ_FOPS(filter_brw_stats); int lproc_filter_attach_seqstat(struct obd_device *dev) { @@ -384,7 +456,60 @@ int lproc_filter_attach_seqstat(struct obd_device *dev) &filter_brw_stats_fops, dev); } +static int filter_per_export_stats_seq_show(struct seq_file *seq, void *v) +{ + struct filter_export_data *fed = seq->private; + + brw_stats_show(seq, &fed->fed_brw_stats); + + return 0; +} + +static ssize_t filter_per_export_stats_seq_write(struct file *file, + const char *buf, size_t len, loff_t *off) +{ + struct seq_file *seq = file->private_data; + struct filter_export_data *fed = seq->private; + int i; + for (i = 0; i < BRW_LAST; i++) + lprocfs_oh_clear(&fed->fed_brw_stats.hist[i]); + + return len; +} + +LPROC_SEQ_FOPS(filter_per_export_stats); + +void lprocfs_filter_init_vars(struct lprocfs_static_vars *lvars) +{ + lvars->module_vars = lprocfs_filter_module_vars; + lvars->obd_vars = lprocfs_filter_obd_vars; +} + +static int filter_per_nid_stats_seq_show(struct seq_file *seq, void *v) +{ + nid_stat_t *tmp = seq->private; + + if (tmp->nid_brw_stats) + brw_stats_show(seq, tmp->nid_brw_stats); + + return 0; +} + +static ssize_t filter_per_nid_stats_seq_write(struct file *file, + const char *buf, size_t len, + loff_t *off) +{ + struct seq_file *seq = file->private_data; + nid_stat_t *tmp = seq->private; + int i; + + if (tmp->nid_brw_stats) + for (i = 0; i < BRW_LAST; i++) + lprocfs_oh_clear(&tmp->nid_brw_stats->hist[i]); + + return len; +} +LPROC_SEQ_FOPS(filter_per_nid_stats); #endif /* LPROCFS */ -LPROCFS_INIT_VARS(filter, lprocfs_module_vars, lprocfs_obd_vars)