diff --git a/doc/man/man1/sacctmgr.1 b/doc/man/man1/sacctmgr.1
index 895e0844ce1a71a03883c23315c60cbf84f47ea2..1293c3f7d11b005e6811da171e9058155effdad3 100644
--- a/doc/man/man1/sacctmgr.1
+++ b/doc/man/man1/sacctmgr.1
@@ -191,6 +191,10 @@ add users or sub accounts to the account they are coordinator over.
This should be a trusted person since they can change limits on
account and user associations inside their realm.
+.TP
+\fIevent\fR
+Events like downed or draining nodes on clusters.
+
.TP
\fIqos\fR
Quality of Service.
@@ -418,15 +422,15 @@ in the \fILIST/SHOW ASSOCIATION FORMAT OPTIONS\fP section.
.SH "SPECIFICATIONS FOR ASSOCIATIONS"
.TP
-\fICluster\fP=<comma separated list of cluster names>
+\fIClusters\fP=<comma separated list of cluster names>
List the associations of the cluster(s).
.TP
-\fIAccount\fP=<comma separated list of account names>
+\fIAccounts\fP=<comma separated list of account names>
List the associations of the account(s).
.TP
-\fIUser\fP=<comma separated list of user names>
+\fIUsers\fP=<comma separated list of user names>
List the associations of the user(s).
.TP
@@ -659,6 +663,133 @@ account or list user.
.RE
+.SH "SPECIFICATIONS FOR EVENTS"
+
+.TP
+\fIAll_Clusters\fP
+Get information on all cluster shortcut.
+
+.TP
+\fIAll_Time\fP
+Get time period for all time shortcut.
+
+.TP
+\fIClusters\fP=<comma separated list of cluster names>
+List the events of the cluster(s). Default is the cluster where the
+command was run.
+
+.TP
+\fIEnd\fP=<OPT>
+Period ending of events. Default is now.
+Valid time formats are...
+HH:MM[:SS] [AM|PM]
+MMDD[YY] or MM/DD[/YY] or MM.DD[.YY]
+MM/DD[/YY]-HH:MM[:SS]
+YYYY-MM-DD[THH:MM[:SS]]
+
+.TP
+\fIEvent\fP=<OPT>
+Specific events to look for, valid options are Cluster or Node,
+default is both.
+
+.TP
+\fIMaxCPUs\fP=<OPT>
+Max number of cpus affected by an event.
+
+.TP
+\fIMinCPUs\fP=<OPT>
+Min number of cpus affected by an event.
+
+.TP
+\fINodes\fP=<comma separated list of node names>
+Node names affected by an event.
+
+.TP
+\fIReason\fP=<comma separated list of reasons>
+Reason an event happened.
+
+.TP
+\fIStart\fP=<OPT>
+Period start of events. Default is 00:00:00 of previous day, unless
+states are given with the States= spec events. If this is the case
+the default behavior is to return events currently in
+the states specified.
+
+Valid time formats are...
+HH:MM[:SS] [AM|PM]
+MMDD[YY] or MM/DD[/YY] or MM.DD[.YY]
+MM/DD[/YY]-HH:MM[:SS]
+YYYY-MM-DD[THH:MM[:SS]]
+
+.TP
+\fIStates\fP=<comma separated list of states>
+State of a node in a node event. If this is set, the event type is
+set automatically to Node.
+
+.TP
+\fIUser\fP=<comma separated list of users>
+Query against users who set the event. If this is set, the event type is
+set automatically to Node since only user slurm can preform a cluster event.
+
+.RE
+
+.SH "LIST/SHOW EVENT FORMAT OPTIONS"
+
+.TP
+\fICluster\fP
+The name of the cluster event happened on.
+
+.TP
+\fIClusterNodes\fP
+The hostlist of nodes on a cluster in a cluster event.
+
+.TP
+\fICPUs\fP
+Number of cpus involved with the event.
+
+.TP
+\fIDuration\fP
+Time period the event was around for.
+
+.TP
+\fIEnd\fP
+Period when event ended.
+
+.TP
+\fIEvent\fP
+Name of the event.
+
+.TP
+\fIEventRaw\fP
+Numeric value of the name of the event.
+
+.TP
+\fINodeName\fP
+The node effected by the event. In a cluster event, this is blank.
+
+.TP
+\fIReason\fP
+The reason an event happened.
+
+.TP
+\fIStart\fP
+Period when event started.
+
+.TP
+\fIState\fP
+On a node event this is the formatted state of the node during the event.
+
+.TP
+\fIStateRaw\fP
+On a node event this is the numeric value of the state of the node
+during the event.
+
+.TP
+\fIUser\fP
+On a node event this is the user who caused the event to happen.
+.RE
+
+
.SH "SPECIFICATIONS FOR QOS"
.TP
diff --git a/src/plugins/accounting_storage/mysql/as_mysql_cluster.c b/src/plugins/accounting_storage/mysql/as_mysql_cluster.c
index cbdb2cdccef378bee1127a7c0f1f8fca889e778d..7c71e37a9f34eac992d3bf59d038ef59e3d54341 100644
--- a/src/plugins/accounting_storage/mysql/as_mysql_cluster.c
+++ b/src/plugins/accounting_storage/mysql/as_mysql_cluster.c
@@ -817,7 +817,7 @@ extern List as_mysql_get_cluster_events(mysql_conn_t *mysql_conn, uint32_t uid,
else
xstrcat(extra, " where (");
- xstrfmtcat(query,
+ xstrfmtcat(extra,
"(time_start < %d) "
"&& (time_end >= %d || time_end = 0))",
event_cond->period_end, event_cond->period_start);
@@ -904,7 +904,9 @@ empty:
if(!(result = mysql_db_query_ret(
mysql_conn->db_conn, query, 0))) {
xfree(query);
- return NULL;
+ list_destroy(ret_list);
+ ret_list = NULL;
+ break;
}
xfree(query);
diff --git a/src/sacctmgr/event_functions.c b/src/sacctmgr/event_functions.c
index 001177389958d4945656526af96ba41e73658040..6865ff7acc22c2a6f2d17582ac2256cbebb3ffe0 100644
--- a/src/sacctmgr/event_functions.c
+++ b/src/sacctmgr/event_functions.c
@@ -41,6 +41,7 @@
#include "src/sacctmgr/sacctmgr.h"
#include "src/common/slurmdbd_defs.h"
#include "src/common/uid.h"
+#include <grp.h>
static uint32_t _decode_node_state(char *val)
{
@@ -166,6 +167,114 @@ static int _addto_state_char_list(List char_list, char *names)
return count;
}
+static char *_convert_to_id(char *name, bool gid)
+{
+ if(gid) {
+ struct group *grp;
+ if (!(grp=getgrnam(name))) {
+ fprintf(stderr, "Invalid group id: %s\n", name);
+ exit(1);
+ }
+ xfree(name);
+ name = xstrdup_printf("%d", grp->gr_gid);
+ } else {
+ struct passwd *pwd;
+ if (!(pwd=getpwnam(name))) {
+ fprintf(stderr, "Invalid user id: %s\n", name);
+ exit(1);
+ }
+ xfree(name);
+ name = xstrdup_printf("%d", pwd->pw_uid);
+ }
+ return name;
+}
+
+/* returns number of objects added to list */
+static int _addto_id_char_list(List char_list, char *names, bool gid)
+{
+ int i=0, start=0;
+ char *name = NULL, *tmp_char = NULL;
+ ListIterator itr = NULL;
+ char quote_c = '\0';
+ int quote = 0;
+ int count = 0;
+
+ if(!char_list) {
+ error("No list was given to fill in");
+ return 0;
+ }
+
+ itr = list_iterator_create(char_list);
+ if(names) {
+ if (names[i] == '\"' || names[i] == '\'') {
+ quote_c = names[i];
+ quote = 1;
+ i++;
+ }
+ start = i;
+ while(names[i]) {
+ //info("got %d - %d = %d", i, start, i-start);
+ if(quote && names[i] == quote_c)
+ break;
+ else if (names[i] == '\"' || names[i] == '\'')
+ names[i] = '`';
+ else if(names[i] == ',') {
+ if((i-start) > 0) {
+ name = xmalloc((i-start+1));
+ memcpy(name, names+start, (i-start));
+ //info("got %s %d", name, i-start);
+ if (!isdigit((int) *name)) {
+ name = _convert_to_id(
+ name, gid);
+ }
+
+ while((tmp_char = list_next(itr))) {
+ if(!strcasecmp(tmp_char, name))
+ break;
+ }
+
+ if(!tmp_char) {
+ list_append(char_list, name);
+ count++;
+ } else
+ xfree(name);
+ list_iterator_reset(itr);
+ }
+ i++;
+ start = i;
+ if(!names[i]) {
+ info("There is a problem with "
+ "your request. It appears you "
+ "have spaces inside your list.");
+ break;
+ }
+ }
+ i++;
+ }
+ if((i-start) > 0) {
+ name = xmalloc((i-start)+1);
+ memcpy(name, names+start, (i-start));
+
+ if (!isdigit((int) *name)) {
+ name = _convert_to_id(name, gid);
+ }
+
+ while((tmp_char = list_next(itr))) {
+ if(!strcasecmp(tmp_char, name))
+ break;
+ }
+
+ if(!tmp_char) {
+ list_append(char_list, name);
+ count++;
+ } else
+ xfree(name);
+ }
+ }
+ list_iterator_destroy(itr);
+ return count;
+}
+
static int _set_cond(int *start, int argc, char *argv[],
slurmdb_event_cond_t *event_cond,
List format_list)
@@ -175,6 +284,7 @@ static int _set_cond(int *start, int argc, char *argv[],
int command_len = 0;
int option = 0;
int local_cluster_flag = 0;
+ int all_time_flag = 0;
if(!event_cond->cluster_list)
event_cond->cluster_list = list_create(slurm_destroy_char);
@@ -191,8 +301,11 @@ static int _set_cond(int *start, int argc, char *argv[],
}
if(!end && !strncasecmp(argv[i], "all_clusters",
- MAX(command_len, 1))) {
+ MAX(command_len, 5))) {
local_cluster_flag = 1;
+ } else if(!end && !strncasecmp(argv[i], "all_time",
+ MAX(command_len, 5))) {
+ all_time_flag = 1;
} else if(!end && !strncasecmp(argv[i], "where",
MAX(command_len, 5))) {
continue;
@@ -245,6 +358,13 @@ static int _set_cond(int *start, int argc, char *argv[],
if(slurm_addto_char_list(event_cond->cluster_list,
argv[i]+end))
set = 1;
+ } else if (!strncasecmp (argv[i], "End", MAX(command_len, 1))) {
+ event_cond->period_end = parse_time(argv[i]+end, 1);
+ set = 1;
+ } else if (!strncasecmp (argv[i], "Format",
+ MAX(command_len, 1))) {
+ if(format_list)
+ slurm_addto_char_list(format_list, argv[i]+end);
} else if (!strncasecmp (argv[i], "MinCpus",
MAX(command_len, 2))) {
if (get_uint(argv[i]+end, &event_cond->cpus_min,
@@ -263,13 +383,6 @@ static int _set_cond(int *start, int argc, char *argv[],
if(slurm_addto_char_list(event_cond->node_list,
argv[i]+end))
set = 1;
- } else if (!strncasecmp (argv[i], "End", MAX(command_len, 1))) {
- event_cond->period_end = parse_time(argv[i]+end, 1);
- set = 1;
- } else if (!strncasecmp (argv[i], "Format",
- MAX(command_len, 1))) {
- if(format_list)
- slurm_addto_char_list(format_list, argv[i]+end);
} else if (!strncasecmp (argv[i], "Reason",
MAX(command_len, 1))) {
if(!event_cond->reason_list)
@@ -288,16 +401,20 @@ static int _set_cond(int *start, int argc, char *argv[],
event_cond->state_list =
list_create(slurm_destroy_char);
if(_addto_state_char_list(event_cond->state_list,
- argv[i]+end))
+ argv[i]+end)) {
+ event_cond->event_type = SLURMDB_EVENT_NODE;
set = 1;
+ }
} else if (!strncasecmp (argv[i], "User",
MAX(command_len, 1))) {
if(!event_cond->reason_uid_list)
event_cond->reason_uid_list =
list_create(slurm_destroy_char);
- if(slurm_addto_char_list(event_cond->reason_uid_list,
- argv[i]+end))
+ if(_addto_id_char_list(event_cond->reason_uid_list,
+ argv[i]+end, 0)) {
+ event_cond->event_type = SLURMDB_EVENT_NODE;
set = 1;
+ }
} else {
exit_code=1;
fprintf(stderr, " Unknown condition: %s\n", argv[i]);
@@ -311,18 +428,26 @@ static int _set_cond(int *start, int argc, char *argv[],
list_append(event_cond->cluster_list, temp);
}
- /* UNCOMMENT THIS IN IF WE EVER DECIDE TO LIMIT THE NUMBER OF EVENTS
- RETURNED */
-
- /* This needs to be done on some systems to make sure
- assoc_cond isn't messed. This has happened on some 64
- bit machines and this is here to be on the safe side.
- */
- /* start_time = event_cond->usage_start; */
- /* end_time = event_cond->usage_end; */
- /* slurmdb_report_set_start_end_time(&event_time, &event_time); */
- /* event_cond->usage_start = start_time; */
- /* event_cond->usage_end = end_time; */
+ if(!all_time_flag && !event_cond->period_start) {
+ event_cond->period_start = time(NULL);
+ if(!event_cond->state_list) {
+ struct tm start_tm;
+
+ if(!localtime_r(&event_cond->period_start, &start_tm)) {
+ fprintf(stderr,
+ " Couldn't get localtime from %ld",
+ event_cond->period_start);
+ exit_code=1;
+ return 0;
+ }
+ start_tm.tm_sec = 0;
+ start_tm.tm_min = 0;
+ start_tm.tm_hour = 0;
+ start_tm.tm_mday--;
+ start_tm.tm_isdst = -1;
+ event_cond->period_start = mktime(&start_tm);
+ }
+ }
return set;
}
@@ -357,6 +482,7 @@ extern int sacctmgr_list_event(int argc, char *argv[])
PRINT_NODENAME,
PRINT_START,
PRINT_REASON,
+ PRINT_STATERAW,
PRINT_STATE,
PRINT_USER
};
@@ -404,7 +530,7 @@ extern int sacctmgr_list_event(int argc, char *argv[])
field = xmalloc(sizeof(print_field_t));
if(!strncasecmp("ClusterNodes", object,
MAX(command_len, 8))) {
- field->type = PRINT_CLUSTER;
+ field->type = PRINT_CLUSTER_NODES;
field->name = xstrdup("Cluster Nodes");
field->len = 20;
field->print_routine = print_fields_str;
@@ -424,7 +550,7 @@ extern int sacctmgr_list_event(int argc, char *argv[])
MAX(command_len, 2))) {
field->type = PRINT_DURATION;
field->name = xstrdup("Duration");
- field->len = 19;
+ field->len = 13;
field->print_routine = print_fields_time_from_secs;
} else if(!strncasecmp("End", object, MAX(command_len, 2))) {
field->type = PRINT_END;
@@ -447,7 +573,12 @@ extern int sacctmgr_list_event(int argc, char *argv[])
MAX(command_len, 1))) {
field->type = PRINT_NODENAME;
field->name = xstrdup("Node Name");
- field->len = 15;
+ field->len = -15;
+ field->print_routine = print_fields_str;
+ } else if(!strncasecmp("Reason", object, MAX(command_len, 1))) {
+ field->type = PRINT_REASON;
+ field->name = xstrdup("Reason");
+ field->len = 30;
field->print_routine = print_fields_str;
} else if(!strncasecmp("Start", object,
MAX(command_len, 1))) {
@@ -455,11 +586,12 @@ extern int sacctmgr_list_event(int argc, char *argv[])
field->name = xstrdup("Start");
field->len = 19;
field->print_routine = print_fields_date;
- } else if(!strncasecmp("Reason", object, MAX(command_len, 1))) {
- field->type = PRINT_REASON;
- field->name = xstrdup("Reason");
- field->len = 20;
- field->print_routine = print_fields_str;
+ } else if(!strncasecmp("StateRaw", object,
+ MAX(command_len, 6))) {
+ field->type = PRINT_STATERAW;
+ field->name = xstrdup("StateRaw");
+ field->len = 8;
+ field->print_routine = print_fields_uint;
} else if(!strncasecmp("State", object, MAX(command_len, 1))) {
field->type = PRINT_STATE;
field->name = xstrdup("State");
@@ -468,7 +600,7 @@ extern int sacctmgr_list_event(int argc, char *argv[])
} else if(!strncasecmp("User", object, MAX(command_len, 1))) {
field->type = PRINT_USER;
field->name = xstrdup("User");
- field->len = 10;
+ field->len = 15;
field->print_routine = print_fields_str;
} else {
exit_code=1;
@@ -571,6 +703,10 @@ extern int sacctmgr_list_event(int argc, char *argv[])
field->print_routine(field, event->reason,
(curr_inx == field_count));
break;
+ case PRINT_STATERAW:
+ field->print_routine(field, event->state,
+ (curr_inx == field_count));
+ break;
case PRINT_STATE:
if(event->event_type == SLURMDB_EVENT_CLUSTER)
tmp_char = NULL;
@@ -583,10 +719,14 @@ extern int sacctmgr_list_event(int argc, char *argv[])
(curr_inx == field_count));
break;
case PRINT_USER:
- tmp_char = uid_to_string(event->reason_uid);
- snprintf(tmp, sizeof(tmp), "%s(%u)",
- tmp_char, event->reason_uid);
- xfree(tmp_char);
+ if(event->reason_uid != NO_VAL) {
+ tmp_char = uid_to_string(
+ event->reason_uid);
+ snprintf(tmp, sizeof(tmp), "%s(%u)",
+ tmp_char, event->reason_uid);
+ xfree(tmp_char);
+ } else
+ memset(tmp, 0, sizeof(tmp));
field->print_routine(field, tmp,
(curr_inx == field_count));
break;
diff --git a/src/sacctmgr/sacctmgr.c b/src/sacctmgr/sacctmgr.c
index 82a34fe6da7aff727bedd16601dc27e5b6357ba8..d442a58e348c84675771c8c044931027075408d5 100644
--- a/src/sacctmgr/sacctmgr.c
+++ b/src/sacctmgr/sacctmgr.c
@@ -851,6 +851,10 @@ sacctmgr [<OPTION>] [<COMMAND>] \n\
add coordinator - Accounts=, and Names= \n\
delete coordinator - Accounts=, and Names= \n\
\n\
+ list events - All_Clusters, All_Time, Clusters=, End=, Events=,\n\
+ Format=, MaxCpus=, MinCpus=, Nodes=, Reason=, \n\
+ Start=, States=, and User= \n\
+ \n\
list qos - Descriptions=, Format=, Ids=, Names=, \n\
and WithDeleted \n\
add qos - Description=, GrpCPUMins=, GrpCPUs=, GrpJobs=, \n\
@@ -916,6 +920,10 @@ sacctmgr [<OPTION>] [<COMMAND>] \n\
MaxCPUs, MaxJobs, MaxNodes, MaxSubmitJobs, \n\
MaxWall, NodeCount, NodeNames \n\
\n\
+ Event - Cluster, ClusterNodes, CPUs, Duration, End, \n\
+ Event, EventRaw, NodeName, Reason, Start, \n\
+ State, StateRaw, User \n\
+ \n\
QOS - Description, GrpCPUMins, GrpCPUs, GrpJobs, \n\
GrpNodes, GrpSubmitJob, GrpWall, ID, \n\
MaxCPUMins, MaxCPUs, MaxJobs, MaxNodes, \n\