4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
27 * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
31 * This file is part of Lustre, http://www.lustre.org/
32 * Lustre is a trademark of Sun Microsystems, Inc.
34 * lnet/selftest/conctl.c
36 * Author: Liang Zhen <liangzhen@clusterfs.com>
41 #include <libcfs/libcfsutil.h>
42 #include <lnet/lnetctl.h>
43 #include <lnet/lnetst.h>
44 /* NB: these includes are layering violation */
45 #include <lustre_ver.h>
46 #include <lustre/lustre_idl.h>
48 lst_sid_t LST_INVALID_SID = {LNET_NID_ANY, -1};
49 static lst_sid_t session_id;
50 static int session_key;
52 #if LUSTRE_VERSION_CODE >= OBD_OCD_VERSION(2, 6, 50, 0)
53 /* assume all nodes can understand feature LST_FEAT_BULK_LEN */
54 static unsigned session_features = LST_FEATS_MASK;
56 static unsigned session_features = LST_FEATS_EMPTY;
59 static lstcon_trans_stat_t trans_stat;
61 typedef struct list_string {
62 struct list_string *lstr_next;
68 # define offsetof(typ,memb) ((unsigned long)((char *)&(((typ *)0)->memb)))
71 static int alloc_count = 0;
72 static int alloc_nob = 0;
77 lstr_t *lstr = malloc(offsetof(lstr_t, lstr_str[sz]));
80 fprintf(stderr, "Can't allocate lstr\n");
87 lstr->lstr_str[0] = 0;
93 free_lstr(lstr_t *lstr)
96 alloc_nob -= lstr->lstr_sz;
101 free_lstrs(lstr_t **list)
105 while ((lstr = *list) != NULL) {
106 *list = lstr->lstr_next;
112 new_lstrs(lstr_t **list, char *prefix, char *postfix,
113 int lo, int hi, int stride)
115 int n1 = strlen(prefix);
116 int n2 = strlen(postfix);
117 int sz = n1 + 20 + n2 + 1;
120 lstr_t *n = alloc_lstr(sz);
122 snprintf(n->lstr_str, sz - 1, "%s%u%s",
123 prefix, lo, postfix);
125 n->lstr_next = *list;
133 expand_lstr(lstr_t **list, lstr_t *l)
135 int nob = strlen(l->lstr_str);
145 b1 = strchr(l->lstr_str, '[');
147 l->lstr_next = *list;
152 b2 = strchr(b1, ']');
153 if (b2 == NULL || b2 == b1 + 1)
161 sep = strchr(expr, ',');
167 if (sscanf(expr, "%u%n", &x, &n) >= 1 && n == nob) {
169 new_lstrs(list, l->lstr_str, b2, x, x, 1);
174 if (sscanf(expr, "%u-%u%n", &x, &y, &n) >= 2 && n == nob &&
177 new_lstrs(list, l->lstr_str, b2, x, y, 1);
182 if (sscanf(expr, "%u-%u/%u%n", &x, &y, &z, &n) >= 3 && n == nob &&
185 new_lstrs(list, l->lstr_str, b2, x, y, z);
191 } while ((expr = sep) != NULL);
199 expand_strs(char *str, lstr_t **head)
207 l = alloc_lstr(strlen(str) + 1);
208 memcpy(l->lstr_str, str, strlen(str) + 1);
216 while ((l = list) != NULL) {
219 rc = expand_lstr(&nlist, l);
221 fprintf(stderr, "Syntax error in \"%s\"\n", str);
229 /* re-order onto 'list' */
230 while ((l = nlist) != NULL) {
231 nlist = l->lstr_next;
236 } while (expanded && rc > 0);
243 while ((l = list) != NULL) {
252 lst_parse_nids(char *str, int *countp, lnet_process_id_t **idspp)
260 rc = expand_strs(str, &head);
270 *idspp = malloc(c * sizeof(lnet_process_id_t));
271 if (*idspp == NULL) {
272 fprintf(stderr, "Out of memory\n");
279 while ((l = head) != NULL) {
283 (*idspp)[i].nid = libcfs_str2nid(l->lstr_str);
284 if ((*idspp)[i].nid == LNET_NID_ANY) {
285 fprintf(stderr, "Invalid nid: %s\n",
290 (*idspp)[i].pid = LUSTRE_LNET_PID;
307 lst_node_state2str(int state)
309 if (state == LST_NODE_ACTIVE)
311 if (state == LST_NODE_BUSY)
313 if (state == LST_NODE_DOWN)
320 lst_node_str2state(char *str)
322 if (strcasecmp(str, "active") == 0)
323 return LST_NODE_ACTIVE;
324 if (strcasecmp(str, "busy") == 0)
325 return LST_NODE_BUSY;
326 if (strcasecmp(str, "down") == 0)
327 return LST_NODE_DOWN;
328 if (strcasecmp(str, "unknown") == 0)
329 return LST_NODE_UNKNOWN;
330 if (strcasecmp(str, "invalid") == 0)
331 return (LST_NODE_UNKNOWN | LST_NODE_DOWN | LST_NODE_BUSY);
337 lst_test_type2name(int type)
339 if (type == LST_TEST_PING)
341 if (type == LST_TEST_BULK)
348 lst_test_name2type(char *name)
350 if (strcasecmp(name, "ping") == 0)
351 return LST_TEST_PING;
352 if (strcasecmp(name, "brw") == 0)
353 return LST_TEST_BULK;
359 lst_print_usage(char *cmd)
361 Parser_printhelp(cmd);
365 lst_print_error(char *sub, const char *def_format, ...)
369 /* local error returned from kernel */
372 fprintf(stderr, "No session exists\n");
375 fprintf(stderr, "Session is shutting down\n");
378 fprintf(stderr, "Unmatched session key or not root\n");
381 fprintf(stderr, "Can't find %s in current session\n", sub);
384 fprintf(stderr, "Invalid parameters list in command line\n");
387 fprintf(stderr, "Bad parameter address\n");
390 fprintf(stderr, "%s already exists\n", sub);
393 va_start(ap, def_format);
394 vfprintf(stderr, def_format, ap);
402 lst_free_rpcent(cfs_list_t *head)
404 lstcon_rpc_ent_t *ent;
406 while (!cfs_list_empty(head)) {
407 ent = cfs_list_entry(head->next, lstcon_rpc_ent_t, rpe_link);
409 cfs_list_del(&ent->rpe_link);
415 lst_reset_rpcent(cfs_list_t *head)
417 lstcon_rpc_ent_t *ent;
419 cfs_list_for_each_entry_typed(ent, head, lstcon_rpc_ent_t, rpe_link) {
420 ent->rpe_sid = LST_INVALID_SID;
421 ent->rpe_peer.nid = LNET_NID_ANY;
422 ent->rpe_peer.pid = LNET_PID_ANY;
423 ent->rpe_rpc_errno = ent->rpe_fwk_errno = 0;
428 lst_alloc_rpcent(cfs_list_t *head, int count, int offset)
430 lstcon_rpc_ent_t *ent;
433 for (i = 0; i < count; i++) {
434 ent = malloc(offsetof(lstcon_rpc_ent_t, rpe_payload[offset]));
436 lst_free_rpcent(head);
440 memset(ent, 0, offsetof(lstcon_rpc_ent_t, rpe_payload[offset]));
442 ent->rpe_sid = LST_INVALID_SID;
443 ent->rpe_peer.nid = LNET_NID_ANY;
444 ent->rpe_peer.pid = LNET_PID_ANY;
445 cfs_list_add(&ent->rpe_link, head);
452 lst_print_transerr(cfs_list_t *head, char *optstr)
454 lstcon_rpc_ent_t *ent;
456 cfs_list_for_each_entry_typed(ent, head, lstcon_rpc_ent_t, rpe_link) {
457 if (ent->rpe_rpc_errno == 0 && ent->rpe_fwk_errno == 0)
460 if (ent->rpe_rpc_errno != 0) {
461 fprintf(stderr, "%s RPC failed on %s: %s\n",
462 optstr, libcfs_id2str(ent->rpe_peer),
463 strerror(ent->rpe_rpc_errno));
467 fprintf(stderr, "operation %s failed on %s: %s\n",
468 optstr, libcfs_id2str(ent->rpe_peer),
469 strerror(ent->rpe_fwk_errno));
473 int lst_info_batch_ioctl(char *batch, int test, int server,
474 lstcon_test_batch_ent_t *entp, int *idxp,
475 int *ndentp, lstcon_node_ent_t *dentsp);
477 int lst_info_group_ioctl(char *name, lstcon_ndlist_ent_t *gent,
478 int *idx, int *count, lstcon_node_ent_t *dents);
480 int lst_query_batch_ioctl(char *batch, int test, int server,
481 int timeout, cfs_list_t *head);
484 lst_ioctl(unsigned int opc, void *buf, int len)
486 struct libcfs_ioctl_data data;
489 LIBCFS_IOC_INIT (data);
490 data.ioc_u32[0] = opc;
491 data.ioc_plen1 = len;
492 data.ioc_pbuf1 = (char *)buf;
493 data.ioc_plen2 = sizeof(trans_stat);
494 data.ioc_pbuf2 = (char *)&trans_stat;
496 memset(&trans_stat, 0, sizeof(trans_stat));
498 rc = l_ioctl(LNET_DEV_ID, IOC_LIBCFS_LNETST, &data);
500 /* local error, no valid RPC result */
505 if (trans_stat.trs_rpc_errno != 0)
508 /* Framework error */
509 if (trans_stat.trs_fwk_errno != 0)
516 lst_new_session_ioctl(char *name, int timeout, int force, lst_sid_t *sid)
518 lstio_session_new_args_t args = {0};
520 args.lstio_ses_key = session_key;
521 args.lstio_ses_timeout = timeout;
522 args.lstio_ses_force = force;
523 args.lstio_ses_idp = sid;
524 args.lstio_ses_feats = session_features;
525 args.lstio_ses_nmlen = strlen(name);
526 args.lstio_ses_namep = name;
528 return lst_ioctl (LSTIO_SESSION_NEW, &args, sizeof(args));
532 jt_lst_new_session(int argc, char **argv)
534 char buf[LST_NAME_SIZE];
542 static struct option session_opts[] =
544 {"timeout", required_argument, 0, 't' },
545 {"force", no_argument, 0, 'f' },
549 if (session_key == 0) {
551 "Can't find env LST_SESSION or value is not valid\n");
557 c = getopt_long(argc, argv, "ft:",
558 session_opts, &optidx);
568 timeout = atoi(optarg);
571 lst_print_usage(argv[0]);
577 fprintf(stderr, "Invalid timeout value\n");
581 if (optind == argc - 1) {
582 name = argv[optind ++];
583 if (strlen(name) >= LST_NAME_SIZE) {
584 fprintf(stderr, "Name size is limited to %d\n",
589 } else if (optind == argc) {
590 char user[LST_NAME_SIZE];
591 char host[LST_NAME_SIZE];
592 struct passwd *pw = getpwuid(getuid());
595 snprintf(user, sizeof(user), "%d", (int)getuid());
597 snprintf(user, sizeof(user), "%s", pw->pw_name);
599 rc = gethostname(host, sizeof(host));
601 snprintf(host, sizeof(host), "unknown_host");
603 snprintf(buf, LST_NAME_SIZE, "%s@%s", user, host);
607 lst_print_usage(argv[0]);
611 rc = lst_new_session_ioctl(name, timeout, force, &session_id);
613 lst_print_error("session", "Failed to create session: %s\n",
618 fprintf(stdout, "SESSION: %s FEATURES: %x TIMEOUT: %d FORCE: %s\n",
619 name, session_features, timeout, force ? "Yes" : "No");
624 lst_session_info_ioctl(char *name, int len, int *key, unsigned *featp,
625 lst_sid_t *sid, lstcon_ndlist_ent_t *ndinfo)
627 lstio_session_info_args_t args = {0};
629 args.lstio_ses_idp = sid;
630 args.lstio_ses_keyp = key;
631 args.lstio_ses_featp = featp;
632 args.lstio_ses_ndinfo = ndinfo;
633 args.lstio_ses_nmlen = len;
634 args.lstio_ses_namep = name;
636 return lst_ioctl(LSTIO_SESSION_INFO, &args, sizeof(args));
640 jt_lst_show_session(int argc, char **argv)
642 lstcon_ndlist_ent_t ndinfo;
644 char name[LST_NAME_SIZE];
649 rc = lst_session_info_ioctl(name, LST_NAME_SIZE, &key,
650 &feats, &sid, &ndinfo);
653 lst_print_error("session", "Failed to show session: %s\n",
658 fprintf(stdout, "%s ID: "LPU64"@%s, KEY: %d FEATURES: %x NODES: %d\n",
659 name, sid.ses_stamp, libcfs_nid2str(sid.ses_nid),
660 key, feats, ndinfo.nle_nnode);
666 lst_end_session_ioctl(void)
668 lstio_session_end_args_t args = {0};
670 args.lstio_ses_key = session_key;
671 return lst_ioctl (LSTIO_SESSION_END, &args, sizeof(args));
675 jt_lst_end_session(int argc, char **argv)
679 if (session_key == 0) {
681 "Can't find env LST_SESSION or value is not valid\n");
685 rc = lst_end_session_ioctl();
688 fprintf(stdout, "session is ended\n");
693 lst_print_error("session", "Failed to end session: %s\n",
698 if (trans_stat.trs_rpc_errno != 0) {
700 "[RPC] Failed to send %d session RPCs: %s\n",
701 lstcon_rpc_stat_failure(&trans_stat, 0),
702 strerror(trans_stat.trs_rpc_errno));
705 if (trans_stat.trs_fwk_errno != 0) {
707 "[FWK] Failed to end session on %d nodes: %s\n",
708 lstcon_sesop_stat_failure(&trans_stat, 0),
709 strerror(trans_stat.trs_fwk_errno));
716 lst_ping_ioctl(char *str, int type, int timeout,
717 int count, lnet_process_id_t *ids, cfs_list_t *head)
719 lstio_debug_args_t args = {0};
721 args.lstio_dbg_key = session_key;
722 args.lstio_dbg_type = type;
723 args.lstio_dbg_flags = 0;
724 args.lstio_dbg_timeout = timeout;
725 args.lstio_dbg_nmlen = (str == NULL) ? 0: strlen(str);
726 args.lstio_dbg_namep = str;
727 args.lstio_dbg_count = count;
728 args.lstio_dbg_idsp = ids;
729 args.lstio_dbg_resultp = head;
731 return lst_ioctl (LSTIO_DEBUG, &args, sizeof(args));
735 lst_get_node_count(int type, char *str, int *countp, lnet_process_id_t **idspp)
737 char buf[LST_NAME_SIZE];
738 lstcon_test_batch_ent_t ent;
739 lstcon_ndlist_ent_t *entp = &ent.tbe_cli_nle;
746 case LST_OPC_SESSION:
747 rc = lst_session_info_ioctl(buf, LST_NAME_SIZE,
748 &key, &feats, &sid, entp);
751 case LST_OPC_BATCHSRV:
752 entp = &ent.tbe_srv_nle;
753 case LST_OPC_BATCHCLI:
754 rc = lst_info_batch_ioctl(str, 0, 0, &ent, NULL, NULL, NULL);
758 rc = lst_info_group_ioctl(str, entp, NULL, NULL, NULL);
762 rc = lst_parse_nids(str, &entp->nle_nnode, idspp) < 0 ? -1 : 0;
771 *countp = entp->nle_nnode;
777 jt_lst_ping(int argc, char **argv)
780 lnet_process_id_t *ids = NULL;
781 lstcon_rpc_ent_t *ent = NULL;
791 static struct option ping_opts[] =
793 {"session", no_argument, 0, 's' },
794 {"server", no_argument, 0, 'v' },
795 {"batch", required_argument, 0, 'b' },
796 {"group", required_argument, 0, 'g' },
797 {"nodes", required_argument, 0, 'n' },
798 {"timeout", required_argument, 0, 't' },
802 if (session_key == 0) {
804 "Can't find env LST_SESSION or value is not valid\n");
810 c = getopt_long(argc, argv, "g:b:n:t:sv",
818 type = LST_OPC_SESSION;
822 type = LST_OPC_GROUP;
827 type = LST_OPC_BATCHCLI;
832 type = LST_OPC_NODES;
837 timeout = atoi(optarg);
845 lst_print_usage(argv[0]);
850 if (type == 0 || timeout <= 0 || optind != argc) {
851 lst_print_usage(argv[0]);
855 if (type == LST_OPC_BATCHCLI && server)
856 type = LST_OPC_BATCHSRV;
858 rc = lst_get_node_count(type, str, &count, &ids);
860 fprintf(stderr, "Failed to get count of nodes from %s: %s\n",
861 (str == NULL) ? "session" : str, strerror(errno));
865 CFS_INIT_LIST_HEAD(&head);
867 rc = lst_alloc_rpcent(&head, count, LST_NAME_SIZE);
869 fprintf(stderr, "Out of memory\n");
874 fprintf(stdout, "Target %s is empty\n",
875 (str == NULL) ? "session" : str);
879 rc = lst_ping_ioctl(str, type, timeout, count, ids, &head);
880 if (rc == -1) { /* local failure */
881 lst_print_error("debug", "Failed to ping %s: %s\n",
882 (str == NULL) ? "session" : str,
888 /* ignore RPC errors and framwork errors */
889 cfs_list_for_each_entry_typed(ent, &head, lstcon_rpc_ent_t, rpe_link) {
890 fprintf(stdout, "\t%s: %s [session: %s id: %s]\n",
891 libcfs_id2str(ent->rpe_peer),
892 lst_node_state2str(ent->rpe_state),
893 (ent->rpe_state == LST_NODE_ACTIVE ||
894 ent->rpe_state == LST_NODE_BUSY)?
895 (ent->rpe_rpc_errno == 0 ?
896 &ent->rpe_payload[0] : "Unknown") :
897 "<NULL>", libcfs_nid2str(ent->rpe_sid.ses_nid));
901 lst_free_rpcent(&head);
911 lst_add_nodes_ioctl (char *name, int count, lnet_process_id_t *ids,
912 unsigned *featp, cfs_list_t *resultp)
914 lstio_group_nodes_args_t args = {0};
916 args.lstio_grp_key = session_key;
917 args.lstio_grp_nmlen = strlen(name);
918 args.lstio_grp_namep = name;
919 args.lstio_grp_count = count;
920 args.lstio_grp_featp = featp;
921 args.lstio_grp_idsp = ids;
922 args.lstio_grp_resultp = resultp;
924 return lst_ioctl(LSTIO_NODES_ADD, &args, sizeof(args));
928 lst_add_group_ioctl (char *name)
930 lstio_group_add_args_t args = {0};
932 args.lstio_grp_key = session_key;
933 args.lstio_grp_nmlen = strlen(name);
934 args.lstio_grp_namep = name;
936 return lst_ioctl(LSTIO_GROUP_ADD, &args, sizeof(args));
940 jt_lst_add_group(int argc, char **argv)
943 lnet_process_id_t *ids;
945 unsigned feats = session_features;
950 if (session_key == 0) {
952 "Can't find env LST_SESSION or value is not valid\n");
957 lst_print_usage(argv[0]);
962 if (strlen(name) >= LST_NAME_SIZE) {
963 fprintf(stderr, "Name length is limited to %d\n",
968 rc = lst_add_group_ioctl(name);
970 lst_print_error("group", "Failed to add group %s: %s\n",
971 name, strerror(errno));
975 CFS_INIT_LIST_HEAD(&head);
977 for (i = 2; i < argc; i++) {
978 /* parse address list */
979 rc = lst_parse_nids(argv[i], &count, &ids);
981 fprintf(stderr, "Ignore invalid id list %s\n",
989 rc = lst_alloc_rpcent(&head, count, 0);
991 fprintf(stderr, "Out of memory\n");
995 rc = lst_add_nodes_ioctl(name, count, ids, &feats, &head);
1002 fprintf(stdout, "%s are added to session\n", argv[i]);
1004 if ((feats & session_features) != session_features) {
1006 "Warning, this session will run with "
1007 "compatible mode because some test nodes "
1008 "might not understand these features: %x\n",
1009 (~feats & session_features));
1012 lst_free_rpcent(&head);
1019 lst_print_error("group", "Failed to add nodes %s: %s\n",
1020 argv[i], strerror(errno));
1023 if (trans_stat.trs_fwk_errno == EPROTO) {
1025 "test nodes might have different LST "
1026 "features, please disable some features by "
1027 "setting LST_FEATURES\n");
1030 lst_print_transerr(&head, "create session");
1033 lst_free_rpcent(&head);
1039 lst_del_group_ioctl (char *name)
1041 lstio_group_del_args_t args = {0};
1043 args.lstio_grp_key = session_key;
1044 args.lstio_grp_nmlen = strlen(name);
1045 args.lstio_grp_namep = name;
1047 return lst_ioctl(LSTIO_GROUP_DEL, &args, sizeof(args));
1051 jt_lst_del_group(int argc, char **argv)
1055 if (session_key == 0) {
1057 "Can't find env LST_SESSION or value is not valid\n");
1062 lst_print_usage(argv[0]);
1066 rc = lst_del_group_ioctl(argv[1]);
1068 fprintf(stdout, "Group is deleted\n");
1073 lst_print_error("group", "Failed to delete group: %s\n",
1078 fprintf(stderr, "Group is deleted with some errors\n");
1080 if (trans_stat.trs_rpc_errno != 0) {
1081 fprintf(stderr, "[RPC] Failed to send %d end session RPCs: %s\n",
1082 lstcon_rpc_stat_failure(&trans_stat, 0),
1083 strerror(trans_stat.trs_rpc_errno));
1086 if (trans_stat.trs_fwk_errno != 0) {
1088 "[FWK] Failed to end session on %d nodes: %s\n",
1089 lstcon_sesop_stat_failure(&trans_stat, 0),
1090 strerror(trans_stat.trs_fwk_errno));
1097 lst_update_group_ioctl(int opc, char *name, int clean, int count,
1098 lnet_process_id_t *ids, cfs_list_t *resultp)
1100 lstio_group_update_args_t args = {0};
1102 args.lstio_grp_key = session_key;
1103 args.lstio_grp_opc = opc;
1104 args.lstio_grp_args = clean;
1105 args.lstio_grp_nmlen = strlen(name);
1106 args.lstio_grp_namep = name;
1107 args.lstio_grp_count = count;
1108 args.lstio_grp_idsp = ids;
1109 args.lstio_grp_resultp = resultp;
1111 return lst_ioctl(LSTIO_GROUP_UPDATE, &args, sizeof(args));
1115 jt_lst_update_group(int argc, char **argv)
1118 lnet_process_id_t *ids = NULL;
1128 static struct option update_group_opts[] =
1130 {"refresh", no_argument, 0, 'f' },
1131 {"clean", required_argument, 0, 'c' },
1132 {"remove", required_argument, 0, 'r' },
1136 if (session_key == 0) {
1138 "Can't find env LST_SESSION or value is not valid\n");
1143 c = getopt_long(argc, argv, "fc:r:",
1144 update_group_opts, &optidx);
1146 /* Detect the end of the options. */
1153 lst_print_usage(argv[0]);
1156 opc = LST_GROUP_REFRESH;
1161 lst_print_usage(argv[0]);
1164 opc = LST_GROUP_RMND;
1169 clean = lst_node_str2state(optarg);
1170 if (opc != 0 || clean <= 0) {
1171 lst_print_usage(argv[0]);
1174 opc = LST_GROUP_CLEAN;
1178 lst_print_usage(argv[0]);
1183 /* no OPC or group is specified */
1184 if (opc == 0 || optind != argc - 1) {
1185 lst_print_usage(argv[0]);
1191 CFS_INIT_LIST_HEAD(&head);
1193 if (opc == LST_GROUP_RMND || opc == LST_GROUP_REFRESH) {
1194 rc = lst_get_node_count(opc == LST_GROUP_RMND ? LST_OPC_NODES :
1196 opc == LST_GROUP_RMND ? str : grp,
1200 fprintf(stderr, "Can't get count of nodes from %s: %s\n",
1201 opc == LST_GROUP_RMND ? str : grp,
1206 rc = lst_alloc_rpcent(&head, count, 0);
1208 fprintf(stderr, "Out of memory\n");
1215 rc = lst_update_group_ioctl(opc, grp, clean, count, ids, &head);
1221 lst_free_rpcent(&head);
1226 lst_free_rpcent(&head);
1227 lst_print_error("group", "Failed to update group: %s\n",
1232 lst_print_transerr(&head, "Updating group");
1234 lst_free_rpcent(&head);
1240 lst_list_group_ioctl(int len, char *name, int idx)
1242 lstio_group_list_args_t args = {0};
1244 args.lstio_grp_key = session_key;
1245 args.lstio_grp_idx = idx;
1246 args.lstio_grp_nmlen = len;
1247 args.lstio_grp_namep = name;
1249 return lst_ioctl(LSTIO_GROUP_LIST, &args, sizeof(args));
1253 lst_info_group_ioctl(char *name, lstcon_ndlist_ent_t *gent,
1254 int *idx, int *count, lstcon_node_ent_t *dents)
1256 lstio_group_info_args_t args = {0};
1258 args.lstio_grp_key = session_key;
1259 args.lstio_grp_nmlen = strlen(name);
1260 args.lstio_grp_namep = name;
1261 args.lstio_grp_entp = gent;
1262 args.lstio_grp_idxp = idx;
1263 args.lstio_grp_ndentp = count;
1264 args.lstio_grp_dentsp = dents;
1266 return lst_ioctl(LSTIO_GROUP_INFO, &args, sizeof(args));
1270 lst_list_group_all(void)
1272 char name[LST_NAME_SIZE];
1276 /* no group is specified, list name of all groups */
1277 for (i = 0; ; i++) {
1278 rc = lst_list_group_ioctl(LST_NAME_SIZE, name, i);
1280 fprintf(stdout, "%d) %s\n", i + 1, name);
1284 if (errno == ENOENT)
1287 lst_print_error("group", "Failed to list group: %s\n",
1292 fprintf(stdout, "Total %d groups\n", i);
1297 #define LST_NODES_TITLE "\tACTIVE\tBUSY\tDOWN\tUNKNOWN\tTOTAL\n"
1300 jt_lst_list_group(int argc, char **argv)
1302 lstcon_ndlist_ent_t gent;
1303 lstcon_node_ent_t *dents;
1318 static struct option list_group_opts[] =
1320 {"active", no_argument, 0, 'a' },
1321 {"busy", no_argument, 0, 'b' },
1322 {"down", no_argument, 0, 'd' },
1323 {"unknown", no_argument, 0, 'u' },
1324 {"all", no_argument, 0, 'l' },
1328 if (session_key == 0) {
1330 "Can't find env LST_SESSION or value is not valid\n");
1335 c = getopt_long(argc, argv, "abdul",
1336 list_group_opts, &optidx);
1343 verbose = active = 1;
1355 verbose = unknown = 1;
1362 lst_print_usage(argv[0]);
1367 if (optind == argc) {
1368 /* no group is specified, list name of all groups */
1369 rc = lst_list_group_all();
1375 fprintf(stdout, LST_NODES_TITLE);
1377 /* list nodes in specified groups */
1378 for (i = optind; i < argc; i++) {
1379 rc = lst_info_group_ioctl(argv[i], &gent, NULL, NULL, NULL);
1381 if (errno == ENOENT) {
1386 lst_print_error("group", "Failed to list group\n",
1392 fprintf(stdout, "\t%d\t%d\t%d\t%d\t%d\t%s\n",
1393 gent.nle_nactive, gent.nle_nbusy,
1394 gent.nle_ndown, gent.nle_nunknown,
1395 gent.nle_nnode, argv[i]);
1399 fprintf(stdout, "Group [ %s ]\n", argv[i]);
1401 if (gent.nle_nnode == 0) {
1402 fprintf(stdout, "No nodes found [ %s ]\n", argv[i]);
1406 count = gent.nle_nnode;
1408 dents = malloc(count * sizeof(lstcon_node_ent_t));
1409 if (dents == NULL) {
1410 fprintf(stderr, "Failed to malloc: %s\n",
1416 rc = lst_info_group_ioctl(argv[i], &gent, &index, &count, dents);
1418 lst_print_error("group", "Failed to list group: %s\n",
1424 for (j = 0, c = 0; j < count; j++) {
1426 ((active && dents[j].nde_state == LST_NODE_ACTIVE) ||
1427 (busy && dents[j].nde_state == LST_NODE_BUSY) ||
1428 (down && dents[j].nde_state == LST_NODE_DOWN) ||
1429 (unknown && dents[j].nde_state == LST_NODE_UNKNOWN))) {
1431 fprintf(stdout, "\t%s: %s\n",
1432 libcfs_id2str(dents[j].nde_id),
1433 lst_node_state2str(dents[j].nde_state));
1438 fprintf(stdout, "Total %d nodes [ %s ]\n", c, argv[i]);
1447 lst_stat_ioctl (char *name, int count, lnet_process_id_t *idsp,
1448 int timeout, cfs_list_t *resultp)
1450 lstio_stat_args_t args = {0};
1452 args.lstio_sta_key = session_key;
1453 args.lstio_sta_timeout = timeout;
1454 args.lstio_sta_nmlen = strlen(name);
1455 args.lstio_sta_namep = name;
1456 args.lstio_sta_count = count;
1457 args.lstio_sta_idsp = idsp;
1458 args.lstio_sta_resultp = resultp;
1460 return lst_ioctl (LSTIO_STAT_QUERY, &args, sizeof(args));
1464 cfs_list_t srp_link;
1467 lnet_process_id_t *srp_ids;
1468 cfs_list_t srp_result[2];
1469 } lst_stat_req_param_t;
1472 lst_stat_req_param_free(lst_stat_req_param_t *srp)
1476 for (i = 0; i < 2; i++)
1477 lst_free_rpcent(&srp->srp_result[i]);
1479 if (srp->srp_ids != NULL)
1486 lst_stat_req_param_alloc(char *name, lst_stat_req_param_t **srpp, int save_old)
1488 lst_stat_req_param_t *srp = NULL;
1489 int count = save_old ? 2 : 1;
1493 srp = malloc(sizeof(*srp));
1497 memset(srp, 0, sizeof(*srp));
1498 CFS_INIT_LIST_HEAD(&srp->srp_result[0]);
1499 CFS_INIT_LIST_HEAD(&srp->srp_result[1]);
1501 rc = lst_get_node_count(LST_OPC_GROUP, name,
1502 &srp->srp_count, NULL);
1503 if (rc != 0 && errno == ENOENT) {
1504 rc = lst_get_node_count(LST_OPC_NODES, name,
1505 &srp->srp_count, &srp->srp_ids);
1510 "Failed to get count of nodes from %s: %s\n",
1511 name, strerror(errno));
1512 lst_stat_req_param_free(srp);
1517 srp->srp_name = name;
1519 for (i = 0; i < count; i++) {
1520 rc = lst_alloc_rpcent(&srp->srp_result[i], srp->srp_count,
1521 sizeof(sfw_counters_t) +
1522 sizeof(srpc_counters_t) +
1523 sizeof(lnet_counters_t));
1525 fprintf(stderr, "Out of memory\n");
1535 lst_stat_req_param_free(srp);
1543 } lst_srpc_stat_result;
1545 #define LST_LNET_AVG 0
1546 #define LST_LNET_MIN 1
1547 #define LST_LNET_MAX 2
1550 float lnet_avg_sndrate;
1551 float lnet_min_sndrate;
1552 float lnet_max_sndrate;
1553 float lnet_total_sndrate;
1555 float lnet_avg_rcvrate;
1556 float lnet_min_rcvrate;
1557 float lnet_max_rcvrate;
1558 float lnet_total_rcvrate;
1560 float lnet_avg_sndperf;
1561 float lnet_min_sndperf;
1562 float lnet_max_sndperf;
1563 float lnet_total_sndperf;
1565 float lnet_avg_rcvperf;
1566 float lnet_min_rcvperf;
1567 float lnet_max_rcvperf;
1568 float lnet_total_rcvperf;
1570 int lnet_stat_count;
1571 } lst_lnet_stat_result_t;
1573 lst_lnet_stat_result_t lnet_stat_result;
1576 lst_lnet_stat_value(int bw, int send, int off)
1580 p = bw ? &lnet_stat_result.lnet_avg_sndperf :
1581 &lnet_stat_result.lnet_avg_sndrate;
1592 lst_timeval_diff(struct timeval *tv1,
1593 struct timeval *tv2, struct timeval *df)
1595 if (tv1->tv_usec >= tv2->tv_usec) {
1596 df->tv_sec = tv1->tv_sec - tv2->tv_sec;
1597 df->tv_usec = tv1->tv_usec - tv2->tv_usec;
1601 df->tv_sec = tv1->tv_sec - 1 - tv2->tv_sec;
1602 df->tv_usec = tv1->tv_usec + 1000000 - tv2->tv_usec;
1608 lst_cal_lnet_stat(float delta, lnet_counters_t *lnet_new,
1609 lnet_counters_t *lnet_old)
1614 perf = (float)(lnet_new->send_length -
1615 lnet_old->send_length) / (1024 * 1024) / delta;
1616 lnet_stat_result.lnet_total_sndperf += perf;
1618 if (lnet_stat_result.lnet_min_sndperf > perf ||
1619 lnet_stat_result.lnet_min_sndperf == 0)
1620 lnet_stat_result.lnet_min_sndperf = perf;
1622 if (lnet_stat_result.lnet_max_sndperf < perf)
1623 lnet_stat_result.lnet_max_sndperf = perf;
1625 perf = (float)(lnet_new->recv_length -
1626 lnet_old->recv_length) / (1024 * 1024) / delta;
1627 lnet_stat_result.lnet_total_rcvperf += perf;
1629 if (lnet_stat_result.lnet_min_rcvperf > perf ||
1630 lnet_stat_result.lnet_min_rcvperf == 0)
1631 lnet_stat_result.lnet_min_rcvperf = perf;
1633 if (lnet_stat_result.lnet_max_rcvperf < perf)
1634 lnet_stat_result.lnet_max_rcvperf = perf;
1636 rate = (lnet_new->send_count - lnet_old->send_count) / delta;
1637 lnet_stat_result.lnet_total_sndrate += rate;
1639 if (lnet_stat_result.lnet_min_sndrate > rate ||
1640 lnet_stat_result.lnet_min_sndrate == 0)
1641 lnet_stat_result.lnet_min_sndrate = rate;
1643 if (lnet_stat_result.lnet_max_sndrate < rate)
1644 lnet_stat_result.lnet_max_sndrate = rate;
1646 rate = (lnet_new->recv_count - lnet_old->recv_count) / delta;
1647 lnet_stat_result.lnet_total_rcvrate += rate;
1649 if (lnet_stat_result.lnet_min_rcvrate > rate ||
1650 lnet_stat_result.lnet_min_rcvrate == 0)
1651 lnet_stat_result.lnet_min_rcvrate = rate;
1653 if (lnet_stat_result.lnet_max_rcvrate < rate)
1654 lnet_stat_result.lnet_max_rcvrate = rate;
1656 lnet_stat_result.lnet_stat_count ++;
1658 lnet_stat_result.lnet_avg_sndrate = lnet_stat_result.lnet_total_sndrate /
1659 lnet_stat_result.lnet_stat_count;
1660 lnet_stat_result.lnet_avg_rcvrate = lnet_stat_result.lnet_total_rcvrate /
1661 lnet_stat_result.lnet_stat_count;
1663 lnet_stat_result.lnet_avg_sndperf = lnet_stat_result.lnet_total_sndperf /
1664 lnet_stat_result.lnet_stat_count;
1665 lnet_stat_result.lnet_avg_rcvperf = lnet_stat_result.lnet_total_rcvperf /
1666 lnet_stat_result.lnet_stat_count;
1671 lst_print_lnet_stat(char *name, int bwrt, int rdwr, int type)
1680 if (lnet_stat_result.lnet_stat_count == 0)
1683 if (bwrt == 1) /* bw only */
1686 if (bwrt == 2) /* rates only */
1689 if (rdwr == 1) /* recv only */
1692 if (rdwr == 2) /* send only */
1695 for (i = start1; i <= end1; i++) {
1696 fprintf(stdout, "[LNet %s of %s]\n",
1697 i == 0 ? "Rates" : "Bandwidth", name);
1699 for (j = start2; j <= end2; j++) {
1700 fprintf(stdout, "[%c] ", j == 0 ? 'R' : 'W');
1702 if ((type & 1) != 0) {
1703 fprintf(stdout, i == 0 ? "Avg: %-8.0f RPC/s " :
1704 "Avg: %-8.2f MB/s ",
1705 lst_lnet_stat_value(i, j, 0));
1708 if ((type & 2) != 0) {
1709 fprintf(stdout, i == 0 ? "Min: %-8.0f RPC/s " :
1710 "Min: %-8.2f MB/s ",
1711 lst_lnet_stat_value(i, j, 1));
1714 if ((type & 4) != 0) {
1715 fprintf(stdout, i == 0 ? "Max: %-8.0f RPC/s" :
1717 lst_lnet_stat_value(i, j, 2));
1720 fprintf(stdout, "\n");
1726 lst_print_stat(char *name, cfs_list_t *resultp,
1727 int idx, int lnet, int bwrt, int rdwr, int type)
1730 lstcon_rpc_ent_t *new;
1731 lstcon_rpc_ent_t *old;
1732 sfw_counters_t *sfwk_new;
1733 sfw_counters_t *sfwk_old;
1734 srpc_counters_t *srpc_new;
1735 srpc_counters_t *srpc_old;
1736 lnet_counters_t *lnet_new;
1737 lnet_counters_t *lnet_old;
1741 CFS_INIT_LIST_HEAD(&tmp[0]);
1742 CFS_INIT_LIST_HEAD(&tmp[1]);
1744 memset(&lnet_stat_result, 0, sizeof(lnet_stat_result));
1746 while (!cfs_list_empty(&resultp[idx])) {
1747 if (cfs_list_empty(&resultp[1 - idx])) {
1748 fprintf(stderr, "Group is changed, re-run stat\n");
1752 new = cfs_list_entry(resultp[idx].next, lstcon_rpc_ent_t,
1754 old = cfs_list_entry(resultp[1 - idx].next, lstcon_rpc_ent_t,
1757 /* first time get stats result, can't calculate diff */
1758 if (new->rpe_peer.nid == LNET_NID_ANY)
1761 if (new->rpe_peer.nid != old->rpe_peer.nid ||
1762 new->rpe_peer.pid != old->rpe_peer.pid) {
1763 /* Something wrong. i.e, somebody change the group */
1767 cfs_list_del(&new->rpe_link);
1768 cfs_list_add_tail(&new->rpe_link, &tmp[idx]);
1770 cfs_list_del(&old->rpe_link);
1771 cfs_list_add_tail(&old->rpe_link, &tmp[1 - idx]);
1773 if (new->rpe_rpc_errno != 0 || new->rpe_fwk_errno != 0 ||
1774 old->rpe_rpc_errno != 0 || old->rpe_fwk_errno != 0) {
1779 sfwk_new = (sfw_counters_t *)&new->rpe_payload[0];
1780 sfwk_old = (sfw_counters_t *)&old->rpe_payload[0];
1782 srpc_new = (srpc_counters_t *)((char *)sfwk_new + sizeof(*sfwk_new));
1783 srpc_old = (srpc_counters_t *)((char *)sfwk_old + sizeof(*sfwk_old));
1785 lnet_new = (lnet_counters_t *)((char *)srpc_new + sizeof(*srpc_new));
1786 lnet_old = (lnet_counters_t *)((char *)srpc_old + sizeof(*srpc_old));
1788 /* Prior to version 2.3, the running_ms field was a counter for
1789 * the number of running tests. We are looking at this value
1790 * to determine if it is a millisecond timestamep (>= 2.3) or a
1791 * test counter (< 2.3). The number 500 is being used for this
1792 * barrier as the test counter should never get this high, and
1793 * the timestamp should never get this low. */
1795 if (sfwk_new->running_ms > 500) {
1796 /* use the timestamp from the remote node, not our
1797 * rpe_stamp from when we copied up the data out of
1800 delta = (float) (sfwk_new->running_ms -
1801 sfwk_old->running_ms) / 1000;
1805 lst_timeval_diff(&new->rpe_stamp, &old->rpe_stamp, &tv);
1806 delta = tv.tv_sec + (float)tv.tv_usec / 1000000;
1809 if (!lnet) /* TODO */
1812 lst_cal_lnet_stat(delta, lnet_new, lnet_old);
1815 cfs_list_splice(&tmp[idx], &resultp[idx]);
1816 cfs_list_splice(&tmp[1 - idx], &resultp[1 - idx]);
1819 fprintf(stdout, "Failed to stat on %d nodes\n", errcount);
1821 if (!lnet) /* TODO */
1824 lst_print_lnet_stat(name, bwrt, rdwr, type);
1828 jt_lst_stat(int argc, char **argv)
1831 lst_stat_req_param_t *srp;
1834 int timeout = 5; /* default timeout, 5 sec */
1835 int delay = 5; /* default delay, 5 sec */
1836 int count = -1; /* run forever */
1837 int lnet = 1; /* lnet stat by default */
1845 static struct option stat_opts[] =
1847 {"timeout" , required_argument, 0, 't' },
1848 {"delay" , required_argument, 0, 'd' },
1849 {"count" , required_argument, 0, 'o' },
1850 {"lnet" , no_argument, 0, 'l' },
1851 {"rpc" , no_argument, 0, 'c' },
1852 {"bw" , no_argument, 0, 'b' },
1853 {"rate" , no_argument, 0, 'a' },
1854 {"read" , no_argument, 0, 'r' },
1855 {"write" , no_argument, 0, 'w' },
1856 {"avg" , no_argument, 0, 'g' },
1857 {"min" , no_argument, 0, 'n' },
1858 {"max" , no_argument, 0, 'x' },
1862 if (session_key == 0) {
1864 "Can't find env LST_SESSION or value is not valid\n");
1869 c = getopt_long(argc, argv, "t:d:lcbarwgnx", stat_opts, &optidx);
1876 timeout = atoi(optarg);
1879 delay = atoi(optarg);
1882 count = atoi(optarg);
1925 lst_print_usage(argv[0]);
1930 if (optind == argc) {
1931 lst_print_usage(argv[0]);
1935 if (timeout <= 0 || delay <= 0) {
1936 fprintf(stderr, "Invalid timeout or delay value\n");
1941 fprintf(stderr, "Invalid count value\n");
1945 /* extra count to get first data point */
1949 CFS_INIT_LIST_HEAD(&head);
1951 while (optind < argc) {
1952 rc = lst_stat_req_param_alloc(argv[optind++], &srp, 1);
1956 cfs_list_add_tail(&srp->srp_link, &head);
1960 time_t now = time(NULL);
1962 if (now - last < delay) {
1963 sleep(delay - now + last);
1969 cfs_list_for_each_entry_typed(srp, &head, lst_stat_req_param_t,
1971 rc = lst_stat_ioctl(srp->srp_name,
1972 srp->srp_count, srp->srp_ids,
1973 timeout, &srp->srp_result[idx]);
1975 lst_print_error("stat", "Failed to stat %s: %s\n",
1976 srp->srp_name, strerror(errno));
1980 lst_print_stat(srp->srp_name, srp->srp_result,
1981 idx, lnet, bwrt, rdwr, type);
1983 lst_reset_rpcent(&srp->srp_result[1 - idx]);
1990 } while (count == -1 || count > 0);
1993 while (!cfs_list_empty(&head)) {
1994 srp = cfs_list_entry(head.next, lst_stat_req_param_t, srp_link);
1996 cfs_list_del(&srp->srp_link);
1997 lst_stat_req_param_free(srp);
2004 jt_lst_show_error(int argc, char **argv)
2007 lst_stat_req_param_t *srp;
2008 lstcon_rpc_ent_t *ent;
2009 sfw_counters_t *sfwk;
2010 srpc_counters_t *srpc;
2017 static struct option show_error_opts[] =
2019 {"session", no_argument, 0, 's' },
2023 if (session_key == 0) {
2025 "Can't find env LST_SESSION or value is not valid\n");
2030 c = getopt_long(argc, argv, "s", show_error_opts, &optidx);
2041 lst_print_usage(argv[0]);
2046 if (optind == argc) {
2047 lst_print_usage(argv[0]);
2051 CFS_INIT_LIST_HEAD(&head);
2053 while (optind < argc) {
2054 rc = lst_stat_req_param_alloc(argv[optind++], &srp, 0);
2058 cfs_list_add_tail(&srp->srp_link, &head);
2061 cfs_list_for_each_entry_typed(srp, &head, lst_stat_req_param_t,
2063 rc = lst_stat_ioctl(srp->srp_name, srp->srp_count,
2064 srp->srp_ids, 10, &srp->srp_result[0]);
2067 lst_print_error(srp->srp_name, "Failed to show errors of %s: %s\n",
2068 srp->srp_name, strerror(errno));
2072 fprintf(stdout, "%s:\n", srp->srp_name);
2076 cfs_list_for_each_entry_typed(ent, &srp->srp_result[0],
2077 lstcon_rpc_ent_t, rpe_link) {
2078 if (ent->rpe_rpc_errno != 0) {
2080 fprintf(stderr, "RPC failure, can't show error on %s\n",
2081 libcfs_id2str(ent->rpe_peer));
2085 if (ent->rpe_fwk_errno != 0) {
2087 fprintf(stderr, "Framework failure, can't show error on %s\n",
2088 libcfs_id2str(ent->rpe_peer));
2092 sfwk = (sfw_counters_t *)&ent->rpe_payload[0];
2093 srpc = (srpc_counters_t *)((char *)sfwk + sizeof(*sfwk));
2095 if (srpc->errors == 0 &&
2096 sfwk->brw_errors == 0 && sfwk->ping_errors == 0)
2100 sfwk->brw_errors == 0 && sfwk->ping_errors == 0)
2105 fprintf(stderr, "%s: [Session %d brw errors, %d ping errors]%c",
2106 libcfs_id2str(ent->rpe_peer),
2107 sfwk->brw_errors, sfwk->ping_errors,
2108 show_rpc ? ' ' : '\n');
2113 fprintf(stderr, "[RPC: %d errors, %d dropped, %d expired]\n",
2114 srpc->errors, srpc->rpcs_dropped, srpc->rpcs_expired);
2117 fprintf(stdout, "Total %d error nodes in %s\n", ecount, srp->srp_name);
2120 while (!cfs_list_empty(&head)) {
2121 srp = cfs_list_entry(head.next, lst_stat_req_param_t, srp_link);
2123 cfs_list_del(&srp->srp_link);
2124 lst_stat_req_param_free(srp);
2131 lst_add_batch_ioctl (char *name)
2133 lstio_batch_add_args_t args = {0};
2135 args.lstio_bat_key = session_key;
2136 args.lstio_bat_nmlen = strlen(name);
2137 args.lstio_bat_namep = name;
2139 return lst_ioctl (LSTIO_BATCH_ADD, &args, sizeof(args));
2143 jt_lst_add_batch(int argc, char **argv)
2148 if (session_key == 0) {
2150 "Can't find env LST_SESSION or value is not valid\n");
2155 lst_print_usage(argv[0]);
2160 if (strlen(name) >= LST_NAME_SIZE) {
2161 fprintf(stderr, "Name length is limited to %d\n",
2166 rc = lst_add_batch_ioctl(name);
2170 lst_print_error("batch", "Failed to create batch: %s\n",
2177 lst_start_batch_ioctl (char *name, int timeout, cfs_list_t *resultp)
2179 lstio_batch_run_args_t args = {0};
2181 args.lstio_bat_key = session_key;
2182 args.lstio_bat_timeout = timeout;
2183 args.lstio_bat_nmlen = strlen(name);
2184 args.lstio_bat_namep = name;
2185 args.lstio_bat_resultp = resultp;
2187 return lst_ioctl(LSTIO_BATCH_START, &args, sizeof(args));
2191 jt_lst_start_batch(int argc, char **argv)
2201 static struct option start_batch_opts[] =
2203 {"timeout", required_argument, 0, 't' },
2207 if (session_key == 0) {
2209 "Can't find env LST_SESSION or value is not valid\n");
2214 c = getopt_long(argc, argv, "t:",
2215 start_batch_opts, &optidx);
2217 /* Detect the end of the options. */
2223 timeout = atoi(optarg);
2226 lst_print_usage(argv[0]);
2231 if (optind == argc) {
2232 batch = LST_DEFAULT_BATCH;
2234 } else if (optind == argc - 1) {
2235 batch = argv[optind];
2238 lst_print_usage(argv[0]);
2242 rc = lst_get_node_count(LST_OPC_BATCHCLI, batch, &count, NULL);
2244 fprintf(stderr, "Failed to get count of nodes from %s: %s\n",
2245 batch, strerror(errno));
2249 CFS_INIT_LIST_HEAD(&head);
2251 rc = lst_alloc_rpcent(&head, count, 0);
2253 fprintf(stderr, "Out of memory\n");
2257 rc = lst_start_batch_ioctl(batch, timeout, &head);
2260 fprintf(stdout, "%s is running now\n", batch);
2261 lst_free_rpcent(&head);
2266 lst_print_error("batch", "Failed to start batch: %s\n",
2268 lst_free_rpcent(&head);
2272 lst_print_transerr(&head, "Run batch");
2274 lst_free_rpcent(&head);
2280 lst_stop_batch_ioctl(char *name, int force, cfs_list_t *resultp)
2282 lstio_batch_stop_args_t args = {0};
2284 args.lstio_bat_key = session_key;
2285 args.lstio_bat_force = force;
2286 args.lstio_bat_nmlen = strlen(name);
2287 args.lstio_bat_namep = name;
2288 args.lstio_bat_resultp = resultp;
2290 return lst_ioctl(LSTIO_BATCH_STOP, &args, sizeof(args));
2294 jt_lst_stop_batch(int argc, char **argv)
2304 static struct option stop_batch_opts[] =
2306 {"force", no_argument, 0, 'f' },
2310 if (session_key == 0) {
2312 "Can't find env LST_SESSION or value is not valid\n");
2317 c = getopt_long(argc, argv, "f",
2318 stop_batch_opts, &optidx);
2320 /* Detect the end of the options. */
2329 lst_print_usage(argv[0]);
2334 if (optind == argc) {
2335 batch = LST_DEFAULT_BATCH;
2337 } else if (optind == argc - 1) {
2338 batch = argv[optind];
2341 lst_print_usage(argv[0]);
2345 rc = lst_get_node_count(LST_OPC_BATCHCLI, batch, &count, NULL);
2347 fprintf(stderr, "Failed to get count of nodes from %s: %s\n",
2348 batch, strerror(errno));
2352 CFS_INIT_LIST_HEAD(&head);
2354 rc = lst_alloc_rpcent(&head, count, 0);
2356 fprintf(stderr, "Out of memory\n");
2360 rc = lst_stop_batch_ioctl(batch, force, &head);
2365 lst_reset_rpcent(&head);
2367 rc = lst_query_batch_ioctl(batch, 0, 0, 30, &head);
2371 if (lstcon_tsbqry_stat_run(&trans_stat, 0) == 0 &&
2372 lstcon_tsbqry_stat_failure(&trans_stat, 0) == 0)
2375 fprintf(stdout, "%d batch in stopping\n",
2376 lstcon_tsbqry_stat_run(&trans_stat, 0));
2380 fprintf(stdout, "Batch is stopped\n");
2381 lst_free_rpcent(&head);
2386 lst_print_error("batch", "Failed to stop batch: %s\n",
2388 lst_free_rpcent(&head);
2392 lst_print_transerr(&head, "stop batch");
2394 lst_free_rpcent(&head);
2400 lst_list_batch_ioctl(int len, char *name, int index)
2402 lstio_batch_list_args_t args = {0};
2404 args.lstio_bat_key = session_key;
2405 args.lstio_bat_idx = index;
2406 args.lstio_bat_nmlen = len;
2407 args.lstio_bat_namep = name;
2409 return lst_ioctl(LSTIO_BATCH_LIST, &args, sizeof(args));
2413 lst_info_batch_ioctl(char *batch, int test, int server,
2414 lstcon_test_batch_ent_t *entp, int *idxp,
2415 int *ndentp, lstcon_node_ent_t *dentsp)
2417 lstio_batch_info_args_t args = {0};
2419 args.lstio_bat_key = session_key;
2420 args.lstio_bat_nmlen = strlen(batch);
2421 args.lstio_bat_namep = batch;
2422 args.lstio_bat_server = server;
2423 args.lstio_bat_testidx = test;
2424 args.lstio_bat_entp = entp;
2425 args.lstio_bat_idxp = idxp;
2426 args.lstio_bat_ndentp = ndentp;
2427 args.lstio_bat_dentsp = dentsp;
2429 return lst_ioctl(LSTIO_BATCH_INFO, &args, sizeof(args));
2433 lst_list_batch_all(void)
2435 char name[LST_NAME_SIZE];
2439 for (i = 0; ; i++) {
2440 rc = lst_list_batch_ioctl(LST_NAME_SIZE, name, i);
2442 fprintf(stdout, "%d) %s\n", i + 1, name);
2446 if (errno == ENOENT)
2449 lst_print_error("batch", "Failed to list batch: %s\n",
2454 fprintf(stdout, "Total %d batches\n", i);
2460 lst_list_tsb_nodes(char *batch, int test, int server,
2461 int count, int active, int invalid)
2463 lstcon_node_ent_t *dents;
2472 /* verbose list, show nodes in batch or test */
2473 dents = malloc(count * sizeof(lstcon_node_ent_t));
2474 if (dents == NULL) {
2475 fprintf(stdout, "Can't allocate memory\n");
2479 rc = lst_info_batch_ioctl(batch, test, server,
2480 NULL, &index, &count, dents);
2483 lst_print_error((test > 0) ? "test" : "batch",
2484 (test > 0) ? "Failed to query test: %s\n" :
2485 "Failed to query batch: %s\n",
2490 for (i = 0, c = 0; i < count; i++) {
2491 if ((!active && dents[i].nde_state == LST_NODE_ACTIVE) ||
2492 (!invalid && (dents[i].nde_state == LST_NODE_BUSY ||
2493 dents[i].nde_state == LST_NODE_DOWN ||
2494 dents[i].nde_state == LST_NODE_UNKNOWN)))
2497 fprintf(stdout, "\t%s: %s\n",
2498 libcfs_id2str(dents[i].nde_id),
2499 lst_node_state2str(dents[i].nde_state));
2503 fprintf(stdout, "Total %d nodes\n", c);
2510 jt_lst_list_batch(int argc, char **argv)
2512 lstcon_test_batch_ent_t ent;
2515 int verbose = 0; /* list nodes in batch or test */
2524 static struct option list_batch_opts[] =
2526 {"test", required_argument, 0, 't' },
2527 {"invalid", no_argument, 0, 'i' },
2528 {"active", no_argument, 0, 'a' },
2529 {"all", no_argument, 0, 'l' },
2530 {"server", no_argument, 0, 's' },
2534 if (session_key == 0) {
2536 "Can't find env LST_SESSION or value is not valid\n");
2541 c = getopt_long(argc, argv, "ailst:",
2542 list_batch_opts, &optidx);
2549 verbose = active = 1;
2552 verbose = invalid = 1;
2555 verbose = active = invalid = 1;
2561 test = atoi(optarg);
2565 lst_print_usage(argv[0]);
2570 if (optind == argc) {
2571 /* list all batches */
2572 rc = lst_list_batch_all();
2576 if (ntest == 1 && test <= 0) {
2577 fprintf(stderr, "Invalid test id, test id starts from 1\n");
2581 if (optind != argc - 1) {
2582 lst_print_usage(argv[0]);
2586 batch = argv[optind];
2589 /* show detail of specified batch or test */
2590 rc = lst_info_batch_ioctl(batch, test, server,
2591 &ent, NULL, NULL, NULL);
2593 lst_print_error((test > 0) ? "test" : "batch",
2594 (test > 0) ? "Failed to query test: %s\n" :
2595 "Failed to query batch: %s\n",
2601 /* list nodes in test or batch */
2602 rc = lst_list_tsb_nodes(batch, test, server,
2603 server ? ent.tbe_srv_nle.nle_nnode :
2604 ent.tbe_cli_nle.nle_nnode,
2609 /* only show number of hosts in batch or test */
2611 fprintf(stdout, "Batch: %s Tests: %d State: %d\n",
2612 batch, ent.u.tbe_batch.bae_ntest,
2613 ent.u.tbe_batch.bae_state);
2614 ntest = ent.u.tbe_batch.bae_ntest;
2615 test = 1; /* starting from test 1 */
2619 "\tTest %d(%s) (loop: %d, concurrency: %d)\n",
2620 test, lst_test_type2name(ent.u.tbe_test.tse_type),
2621 ent.u.tbe_test.tse_loop,
2622 ent.u.tbe_test.tse_concur);
2627 fprintf(stdout, LST_NODES_TITLE);
2628 fprintf(stdout, "client\t%d\t%d\t%d\t%d\t%d\n"
2629 "server\t%d\t%d\t%d\t%d\t%d\n",
2630 ent.tbe_cli_nle.nle_nactive,
2631 ent.tbe_cli_nle.nle_nbusy,
2632 ent.tbe_cli_nle.nle_ndown,
2633 ent.tbe_cli_nle.nle_nunknown,
2634 ent.tbe_cli_nle.nle_nnode,
2635 ent.tbe_srv_nle.nle_nactive,
2636 ent.tbe_srv_nle.nle_nbusy,
2637 ent.tbe_srv_nle.nle_ndown,
2638 ent.tbe_srv_nle.nle_nunknown,
2639 ent.tbe_srv_nle.nle_nnode);
2648 lst_query_batch_ioctl(char *batch, int test, int server,
2649 int timeout, cfs_list_t *head)
2651 lstio_batch_query_args_t args = {0};
2653 args.lstio_bat_key = session_key;
2654 args.lstio_bat_testidx = test;
2655 args.lstio_bat_client = !(server);
2656 args.lstio_bat_timeout = timeout;
2657 args.lstio_bat_nmlen = strlen(batch);
2658 args.lstio_bat_namep = batch;
2659 args.lstio_bat_resultp = head;
2661 return lst_ioctl(LSTIO_BATCH_QUERY, &args, sizeof(args));
2665 lst_print_tsb_verbose(cfs_list_t *head,
2666 int active, int idle, int error)
2668 lstcon_rpc_ent_t *ent;
2670 cfs_list_for_each_entry_typed(ent, head, lstcon_rpc_ent_t, rpe_link) {
2671 if (ent->rpe_priv[0] == 0 && active)
2674 if (ent->rpe_priv[0] != 0 && idle)
2677 if (ent->rpe_fwk_errno == 0 && error)
2680 fprintf(stdout, "%s [%s]: %s\n",
2681 libcfs_id2str(ent->rpe_peer),
2682 lst_node_state2str(ent->rpe_state),
2683 ent->rpe_rpc_errno != 0 ?
2684 strerror(ent->rpe_rpc_errno) :
2685 (ent->rpe_priv[0] > 0 ? "Running" : "Idle"));
2690 jt_lst_query_batch(int argc, char **argv)
2692 lstcon_test_batch_ent_t ent;
2699 int timeout = 5; /* default 5 seconds */
2700 int delay = 5; /* default 5 seconds */
2701 int loop = 1; /* default 1 loop */
2711 static struct option query_batch_opts[] =
2713 {"timeout", required_argument, 0, 'o' },
2714 {"delay", required_argument, 0, 'd' },
2715 {"loop", required_argument, 0, 'c' },
2716 {"test", required_argument, 0, 't' },
2717 {"server", no_argument, 0, 's' },
2718 {"active", no_argument, 0, 'a' },
2719 {"idle", no_argument, 0, 'i' },
2720 {"error", no_argument, 0, 'e' },
2721 {"all", no_argument, 0, 'l' },
2725 if (session_key == 0) {
2727 "Can't find env LST_SESSION or value is not valid\n");
2732 c = getopt_long(argc, argv, "o:d:c:t:saiel",
2733 query_batch_opts, &optidx);
2735 /* Detect the end of the options. */
2741 timeout = atoi(optarg);
2744 delay = atoi(optarg);
2747 loop = atoi(optarg);
2750 test = atoi(optarg);
2756 active = verbose = 1;
2762 error = verbose = 1;
2768 lst_print_usage(argv[0]);
2773 if (test < 0 || timeout <= 0 || delay <= 0 || loop <= 0) {
2774 lst_print_usage(argv[0]);
2778 if (optind == argc) {
2779 batch = LST_DEFAULT_BATCH;
2781 } else if (optind == argc - 1) {
2782 batch = argv[optind];
2785 lst_print_usage(argv[0]);
2790 CFS_INIT_LIST_HEAD(&head);
2793 rc = lst_info_batch_ioctl(batch, test, server,
2794 &ent, NULL, NULL, NULL);
2796 fprintf(stderr, "Failed to query %s [%d]: %s\n",
2797 batch, test, strerror(errno));
2801 count = server ? ent.tbe_srv_nle.nle_nnode :
2802 ent.tbe_cli_nle.nle_nnode;
2804 fprintf(stdout, "Batch or test is empty\n");
2809 rc = lst_alloc_rpcent(&head, count, 0);
2811 fprintf(stderr, "Out of memory\n");
2815 for (i = 0; i < loop; i++) {
2816 time_t now = time(NULL);
2818 if (now - last < delay) {
2819 sleep(delay - now + last);
2825 rc = lst_query_batch_ioctl(batch, test,
2826 server, timeout, &head);
2828 fprintf(stderr, "Failed to query batch: %s\n",
2835 lst_print_tsb_verbose(&head, active, idle, error);
2839 fprintf(stdout, "%s [%d] ", batch, test);
2841 if (lstcon_rpc_stat_failure(&trans_stat, 0) != 0) {
2842 fprintf(stdout, "%d of %d nodes are unknown, ",
2843 lstcon_rpc_stat_failure(&trans_stat, 0),
2844 lstcon_rpc_stat_total(&trans_stat, 0));
2847 if (lstcon_rpc_stat_failure(&trans_stat, 0) == 0 &&
2848 lstcon_tsbqry_stat_run(&trans_stat, 0) == 0 &&
2849 lstcon_tsbqry_stat_failure(&trans_stat, 0) == 0) {
2850 fprintf(stdout, "is stopped\n");
2854 if (lstcon_rpc_stat_failure(&trans_stat, 0) == 0 &&
2855 lstcon_tsbqry_stat_idle(&trans_stat, 0) == 0 &&
2856 lstcon_tsbqry_stat_failure(&trans_stat, 0) == 0) {
2857 fprintf(stdout, "is running\n");
2861 fprintf(stdout, "stopped: %d , running: %d, failed: %d\n",
2862 lstcon_tsbqry_stat_idle(&trans_stat, 0),
2863 lstcon_tsbqry_stat_run(&trans_stat, 0),
2864 lstcon_tsbqry_stat_failure(&trans_stat, 0));
2867 lst_free_rpcent(&head);
2873 lst_parse_distribute(char *dstr, int *dist, int *span)
2879 dstr = strchr(dstr, ':');
2883 *span = atoi(dstr + 1);
2891 lst_get_bulk_param(int argc, char **argv, lst_test_bulk_param_t *bulk)
2898 bulk->blk_size = 4096;
2899 bulk->blk_opc = LST_BRW_READ;
2900 bulk->blk_flags = LST_BRW_CHECK_NONE;
2903 if (strcasestr(argv[i], "check=") == argv[i] ||
2904 strcasestr(argv[i], "c=") == argv[i]) {
2905 tok = strchr(argv[i], '=') + 1;
2907 if (strcasecmp(tok, "full") == 0) {
2908 bulk->blk_flags = LST_BRW_CHECK_FULL;
2909 } else if (strcasecmp(tok, "simple") == 0) {
2910 bulk->blk_flags = LST_BRW_CHECK_SIMPLE;
2912 fprintf(stderr, "Unknow flag %s\n", tok);
2916 } else if (strcasestr(argv[i], "size=") == argv[i] ||
2917 strcasestr(argv[i], "s=") == argv[i]) {
2918 tok = strchr(argv[i], '=') + 1;
2920 bulk->blk_size = strtol(tok, &end, 0);
2921 if (bulk->blk_size <= 0) {
2922 fprintf(stderr, "Invalid size %s\n", tok);
2929 if (*end == 'k' || *end == 'K')
2930 bulk->blk_size *= 1024;
2931 else if (*end == 'm' || *end == 'M')
2932 bulk->blk_size *= 1024 * 1024;
2934 if (bulk->blk_size > CFS_PAGE_SIZE * LNET_MAX_IOV) {
2935 fprintf(stderr, "Size exceed limitation: %d bytes\n",
2940 } else if (strcasecmp(argv[i], "read") == 0 ||
2941 strcasecmp(argv[i], "r") == 0) {
2942 bulk->blk_opc = LST_BRW_READ;
2944 } else if (strcasecmp(argv[i], "write") == 0 ||
2945 strcasecmp(argv[i], "w") == 0) {
2946 bulk->blk_opc = LST_BRW_WRITE;
2949 fprintf(stderr, "Unknow parameter: %s\n", argv[i]);
2960 lst_get_test_param(char *test, int argc, char **argv, void **param, int *plen)
2962 lst_test_bulk_param_t *bulk = NULL;
2965 type = lst_test_name2type(test);
2967 fprintf(stderr, "Unknow test name %s\n", test);
2976 bulk = malloc(sizeof(*bulk));
2978 fprintf(stderr, "Out of memory\n");
2982 memset(bulk, 0, sizeof(*bulk));
2984 if (lst_get_bulk_param(argc, argv, bulk) != 0) {
2990 *plen = sizeof(*bulk);
2998 /* TODO: parse more parameter */
3003 lst_add_test_ioctl(char *batch, int type, int loop, int concur,
3004 int dist, int span, char *sgrp, char *dgrp,
3005 void *param, int plen, int *retp, cfs_list_t *resultp)
3007 lstio_test_args_t args = {0};
3009 args.lstio_tes_key = session_key;
3010 args.lstio_tes_bat_nmlen = strlen(batch);
3011 args.lstio_tes_bat_name = batch;
3012 args.lstio_tes_type = type;
3013 args.lstio_tes_oneside = 0;
3014 args.lstio_tes_loop = loop;
3015 args.lstio_tes_concur = concur;
3016 args.lstio_tes_dist = dist;
3017 args.lstio_tes_span = span;
3018 args.lstio_tes_sgrp_nmlen = strlen(sgrp);
3019 args.lstio_tes_sgrp_name = sgrp;
3020 args.lstio_tes_dgrp_nmlen = strlen(dgrp);
3021 args.lstio_tes_dgrp_name = dgrp;
3022 args.lstio_tes_param_len = plen;
3023 args.lstio_tes_param = param;
3024 args.lstio_tes_retp = retp;
3025 args.lstio_tes_resultp = resultp;
3027 return lst_ioctl(LSTIO_TEST_ADD, &args, sizeof(args));
3031 jt_lst_add_test(int argc, char **argv)
3053 static struct option add_test_opts[] =
3055 {"batch", required_argument, 0, 'b' },
3056 {"concurrency", required_argument, 0, 'c' },
3057 {"distribute", required_argument, 0, 'd' },
3058 {"from", required_argument, 0, 'f' },
3059 {"to", required_argument, 0, 't' },
3060 {"loop", required_argument, 0, 'l' },
3064 if (session_key == 0) {
3066 "Can't find env LST_SESSION or value is not valid\n");
3071 c = getopt_long(argc, argv, "b:c:d:f:l:t:",
3072 add_test_opts, &optidx);
3074 /* Detect the end of the options. */
3083 concur = atoi(optarg);
3092 loop = atoi(optarg);
3098 lst_print_usage(argv[0]);
3103 if (optind == argc || from == NULL || to == NULL) {
3104 lst_print_usage(argv[0]);
3108 if (concur <= 0 || concur > LST_MAX_CONCUR) {
3109 fprintf(stderr, "Invalid concurrency of test: %d\n", concur);
3114 batch = LST_DEFAULT_BATCH;
3117 rc = lst_parse_distribute(dstr, &dist, &span);
3119 fprintf(stderr, "Invalid distribution: %s\n", dstr);
3124 test = argv[optind++];
3129 type = lst_get_test_param(test, argc, argv, ¶m, &plen);
3131 fprintf(stderr, "Failed to add test (%s)\n", test);
3135 CFS_INIT_LIST_HEAD(&head);
3137 rc = lst_get_node_count(LST_OPC_GROUP, from, &fcount, NULL);
3139 fprintf(stderr, "Can't get count of nodes from %s: %s\n",
3140 from, strerror(errno));
3144 rc = lst_get_node_count(LST_OPC_GROUP, to, &tcount, NULL);
3146 fprintf(stderr, "Can't get count of nodes from %s: %s\n",
3147 to, strerror(errno));
3151 rc = lst_alloc_rpcent(&head, fcount > tcount ? fcount : tcount, 0);
3153 fprintf(stderr, "Out of memory\n");
3157 rc = lst_add_test_ioctl(batch, type, loop, concur,
3158 dist, span, from, to, param, plen, &ret, &head);
3161 fprintf(stdout, "Test was added successfully\n");
3163 fprintf(stdout, "Server group contains userland test "
3164 "nodes, old version of tcplnd can't accept "
3165 "connection request\n");
3172 lst_print_error("test", "Failed to add test: %s\n",
3177 lst_print_transerr(&head, "add test");
3179 lst_free_rpcent(&head);
3187 static command_t lst_cmdlist[] = {
3188 {"new_session", jt_lst_new_session, NULL,
3189 "Usage: lst new_session [--timeout TIME] [--force] [NAME]" },
3190 {"end_session", jt_lst_end_session, NULL,
3191 "Usage: lst end_session" },
3192 {"show_session", jt_lst_show_session, NULL,
3193 "Usage: lst show_session" },
3194 {"ping", jt_lst_ping , NULL,
3195 "Usage: lst ping [--group NAME] [--batch NAME] [--session] [--nodes IDS]" },
3196 {"add_group", jt_lst_add_group, NULL,
3197 "Usage: lst group NAME IDs [IDs]..." },
3198 {"del_group", jt_lst_del_group, NULL,
3199 "Usage: lst del_group NAME" },
3200 {"update_group", jt_lst_update_group, NULL,
3201 "Usage: lst update_group NAME [--clean] [--refresh] [--remove IDs]" },
3202 {"list_group", jt_lst_list_group, NULL,
3203 "Usage: lst list_group [--active] [--busy] [--down] [--unknown] GROUP ..." },
3204 {"stat", jt_lst_stat, NULL,
3205 "Usage: lst stat [--bw] [--rate] [--read] [--write] [--max] [--min] [--avg] "
3206 " [--timeout #] [--delay #] [--count #] GROUP [GROUP]" },
3207 {"show_error", jt_lst_show_error, NULL,
3208 "Usage: lst show_error NAME | IDS ..." },
3209 {"add_batch", jt_lst_add_batch, NULL,
3210 "Usage: lst add_batch NAME" },
3211 {"run", jt_lst_start_batch, NULL,
3212 "Usage: lst run [--timeout TIME] [NAME]" },
3213 {"stop", jt_lst_stop_batch, NULL,
3214 "Usage: lst stop [--force] BATCH_NAME" },
3215 {"list_batch", jt_lst_list_batch, NULL,
3216 "Usage: lst list_batch NAME [--test ID] [--server]" },
3217 {"query", jt_lst_query_batch, NULL,
3218 "Usage: lst query [--test ID] [--server] [--timeout TIME] NAME" },
3219 {"add_test", jt_lst_add_test, NULL,
3220 "Usage: lst add_test [--batch BATCH] [--loop #] [--concurrency #] "
3221 " [--distribute #:#] [--from GROUP] [--to GROUP] TEST..." },
3222 {"help", Parser_help, 0, "help" },
3227 lst_initialize(void)
3232 feats = getenv("LST_FEATURES");
3234 session_features = strtol(feats, NULL, 16);
3236 if ((session_features & ~LST_FEATS_MASK) != 0) {
3238 "Unsupported session features %x, "
3239 "only support these features so far: %x\n",
3240 (session_features & ~LST_FEATS_MASK), LST_FEATS_MASK);
3244 key = getenv("LST_SESSION");
3251 session_key = atoi(key);
3257 main(int argc, char **argv)
3263 rc = libcfs_arch_init();
3267 rc = lst_initialize();
3271 rc = ptl_initialize(argc, argv);
3275 Parser_init("lst > ", lst_cmdlist);
3278 rc = Parser_execarg(argc - 1, argv + 1, lst_cmdlist);
3285 libcfs_arch_cleanup();