extern int slurm_ckpt_unpack_job(check_jobinfo_t jobinfo, Buf buffer,
uint16_t protocol_version)
{
uint32_t uint32_tmp;
struct check_job_info *check_ptr =
(struct check_job_info *)jobinfo;
if (protocol_version >= SLURM_MIN_PROTOCOL_VERSION) {
uint16_t id;
uint32_t size;
safe_unpack16(&id, buffer);
safe_unpack32(&size, buffer);
if (id != CHECK_POE) {
uint32_t x;
x = get_buf_offset(buffer);
set_buf_offset(buffer, x + size);
} else {
safe_unpack16(&check_ptr->disabled, buffer);
safe_unpack16(&check_ptr->node_cnt, buffer);
safe_unpack16(&check_ptr->reply_cnt, buffer);
safe_unpack16(&check_ptr->wait_time, buffer);
safe_unpack32(&check_ptr->error_code, buffer);
safe_unpackstr_xmalloc(&check_ptr->error_msg,
&uint32_tmp, buffer);
safe_unpack_time(&check_ptr->time_stamp, buffer);
}
}
return SLURM_SUCCESS;
unpack_error:
xfree(check_ptr->error_msg);
return SLURM_ERROR;
}
extern int slurm_ckpt_unpack_job(check_jobinfo_t jobinfo, Buf buffer,
uint16_t protocol_version)
{
uint32_t uint32_tmp;
struct check_job_info *check_ptr =
(struct check_job_info *)jobinfo;
if (protocol_version >= SLURM_2_3_PROTOCOL_VERSION) {
safe_unpack16(&check_ptr->disabled, buffer);
safe_unpack16(&check_ptr->node_cnt, buffer);
safe_unpack16(&check_ptr->reply_cnt, buffer);
safe_unpack16(&check_ptr->wait_time, buffer);
safe_unpack32(&check_ptr->error_code, buffer);
safe_unpackstr_xmalloc(&check_ptr->error_msg,
&uint32_tmp, buffer);
safe_unpack_time(&check_ptr->time_stamp, buffer);
}
return SLURM_SUCCESS;
unpack_error:
xfree(check_ptr->error_msg);
return SLURM_ERROR;
}
static int _unpack_node_state_msg(dbd_node_state_msg_t **msg,
uint16_t rpc_version, Buf buffer)
{
dbd_node_state_msg_t *msg_ptr;
uint32_t uint32_tmp;
msg_ptr = xmalloc(sizeof(dbd_node_state_msg_t));
*msg = msg_ptr;
msg_ptr->reason_uid = NO_VAL;
if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpackstr_xmalloc(&msg_ptr->hostlist, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->reason, &uint32_tmp, buffer);
safe_unpack32(&msg_ptr->reason_uid, buffer);
safe_unpack16(&msg_ptr->new_state, buffer);
safe_unpack_time(&msg_ptr->event_time, buffer);
safe_unpack32(&msg_ptr->state, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_str,
&uint32_tmp, buffer);
}
return SLURM_SUCCESS;
unpack_error:
slurmdbd_free_node_state_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
extern int acct_gather_energy_unpack(acct_gather_energy_t **energy, Buf buffer,
uint16_t protocol_version)
{
acct_gather_energy_t *energy_ptr = acct_gather_energy_alloc();
*energy = energy_ptr;
if (protocol_version >= SLURM_2_6_PROTOCOL_VERSION) {
safe_unpack32(&energy_ptr->base_consumed_energy, buffer);
safe_unpack32(&energy_ptr->base_watts, buffer);
safe_unpack32(&energy_ptr->consumed_energy, buffer);
safe_unpack32(&energy_ptr->current_watts, buffer);
safe_unpack32(&energy_ptr->previous_consumed_energy, buffer);
safe_unpack_time(&energy_ptr->poll_time, buffer);
} else {
safe_unpack32(&energy_ptr->base_consumed_energy, buffer);
safe_unpack32(&energy_ptr->base_watts, buffer);
safe_unpack32(&energy_ptr->consumed_energy, buffer);
safe_unpack32(&energy_ptr->current_watts, buffer);
}
return SLURM_SUCCESS;
unpack_error:
acct_gather_energy_destroy(energy_ptr);
*energy = NULL;
return SLURM_ERROR;
}
extern int slurmdbd_unpack_usage_msg(dbd_usage_msg_t **msg,
uint16_t rpc_version,
slurmdbd_msg_type_t type,
Buf buffer)
{
dbd_usage_msg_t *msg_ptr = NULL;
int (*my_rec) (void **object, uint16_t rpc_version, Buf buffer);
msg_ptr = xmalloc(sizeof(dbd_usage_msg_t));
*msg = msg_ptr;
switch (type) {
case DBD_GET_ASSOC_USAGE:
case DBD_GOT_ASSOC_USAGE:
my_rec = slurmdb_unpack_assoc_rec;
break;
case DBD_GET_CLUSTER_USAGE:
case DBD_GOT_CLUSTER_USAGE:
my_rec = slurmdb_unpack_cluster_rec;
break;
case DBD_GET_WCKEY_USAGE:
case DBD_GOT_WCKEY_USAGE:
my_rec = slurmdb_unpack_wckey_rec;
break;
default:
fatal("Unknown pack type");
return SLURM_ERROR;
}
if ((*(my_rec))(&msg_ptr->rec, rpc_version, buffer) == SLURM_ERROR)
goto unpack_error;
safe_unpack_time(&msg_ptr->start, buffer);
safe_unpack_time(&msg_ptr->end, buffer);
return SLURM_SUCCESS;
unpack_error:
slurmdbd_free_usage_msg(msg_ptr, type);
*msg = NULL;
return SLURM_ERROR;
}
extern int acct_gather_energy_unpack(acct_gather_energy_t **energy, Buf buffer,
uint16_t protocol_version, bool need_alloc)
{
uint32_t uint32_tmp;
acct_gather_energy_t *energy_ptr;
if (need_alloc) {
energy_ptr = acct_gather_energy_alloc(1);
*energy = energy_ptr;
} else {
energy_ptr = *energy;
}
if (protocol_version >= SLURM_15_08_PROTOCOL_VERSION) {
safe_unpack64(&energy_ptr->base_consumed_energy, buffer);
safe_unpack32(&energy_ptr->base_watts, buffer);
safe_unpack64(&energy_ptr->consumed_energy, buffer);
safe_unpack32(&energy_ptr->current_watts, buffer);
safe_unpack64(&energy_ptr->previous_consumed_energy, buffer);
safe_unpack_time(&energy_ptr->poll_time, buffer);
} else if (protocol_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpack32(&uint32_tmp, buffer);
energy_ptr->base_consumed_energy = (uint64_t) uint32_tmp;
safe_unpack32(&energy_ptr->base_watts, buffer);
safe_unpack32(&uint32_tmp, buffer);
energy_ptr->consumed_energy = (uint64_t) uint32_tmp;
safe_unpack32(&energy_ptr->current_watts, buffer);
safe_unpack32(&uint32_tmp, buffer);
energy_ptr->previous_consumed_energy = (uint64_t) uint32_tmp;
safe_unpack_time(&energy_ptr->poll_time, buffer);
}
return SLURM_SUCCESS;
unpack_error:
if (need_alloc) {
acct_gather_energy_destroy(energy_ptr);
*energy = NULL;
} else
memset(energy_ptr, 0, sizeof(acct_gather_energy_t));
return SLURM_ERROR;
}
static int _unpack_roll_usage_msg(dbd_roll_usage_msg_t **msg,
uint16_t rpc_version, Buf buffer)
{
dbd_roll_usage_msg_t *msg_ptr = xmalloc(sizeof(dbd_roll_usage_msg_t));
*msg = msg_ptr;
if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpack16(&msg_ptr->archive_data, buffer);
safe_unpack_time(&msg_ptr->end, buffer);
safe_unpack_time(&msg_ptr->start, buffer);
}
return SLURM_SUCCESS;
unpack_error:
slurmdbd_free_roll_usage_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
static int _unpack_job_suspend_msg(dbd_job_suspend_msg_t **msg,
uint16_t rpc_version, Buf buffer)
{
dbd_job_suspend_msg_t *msg_ptr = xmalloc(sizeof(dbd_job_suspend_msg_t));
*msg = msg_ptr;
if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpack32(&msg_ptr->assoc_id, buffer);
safe_unpack64(&msg_ptr->db_index, buffer);
safe_unpack32(&msg_ptr->job_id, buffer);
safe_unpack32(&msg_ptr->job_state, buffer);
safe_unpack_time(&msg_ptr->submit_time, buffer);
safe_unpack_time(&msg_ptr->suspend_time, buffer);
}
return SLURM_SUCCESS;
unpack_error:
slurmdbd_free_job_suspend_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
static int _unpack_step_start_msg(dbd_step_start_msg_t **msg,
uint16_t rpc_version, Buf buffer)
{
uint32_t uint32_tmp = 0;
dbd_step_start_msg_t *msg_ptr = xmalloc(sizeof(dbd_step_start_msg_t));
*msg = msg_ptr;
if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpack32(&msg_ptr->assoc_id, buffer);
safe_unpack64(&msg_ptr->db_index, buffer);
safe_unpack32(&msg_ptr->job_id, buffer);
safe_unpackstr_xmalloc(&msg_ptr->name, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->nodes, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->node_inx, &uint32_tmp, buffer);
safe_unpack32(&msg_ptr->node_cnt, buffer);
safe_unpack_time(&msg_ptr->start_time, buffer);
safe_unpack_time(&msg_ptr->job_submit_time, buffer);
safe_unpack32(&msg_ptr->req_cpufreq_min, buffer);
safe_unpack32(&msg_ptr->req_cpufreq_max, buffer);
safe_unpack32(&msg_ptr->req_cpufreq_gov, buffer);
safe_unpack32(&msg_ptr->step_id, buffer);
safe_unpack32(&msg_ptr->task_dist, buffer);
safe_unpack32(&msg_ptr->total_tasks, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_alloc_str,
&uint32_tmp, buffer);
} else
goto unpack_error;
return SLURM_SUCCESS;
unpack_error:
debug2("slurmdbd_unpack_step_start_msg:"
"unpack_error: size_buf(buffer) %u",
size_buf(buffer));
slurmdbd_free_step_start_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
static int _unpack_job_complete_msg(dbd_job_comp_msg_t **msg,
uint16_t rpc_version, Buf buffer)
{
uint32_t uint32_tmp;
dbd_job_comp_msg_t *msg_ptr = xmalloc(sizeof(dbd_job_comp_msg_t));
*msg = msg_ptr;
if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpackstr_xmalloc(&msg_ptr->admin_comment,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->assoc_id, buffer);
safe_unpackstr_xmalloc(&msg_ptr->comment, &uint32_tmp, buffer);
safe_unpack64(&msg_ptr->db_index, buffer);
safe_unpack32(&msg_ptr->derived_ec, buffer);
safe_unpack_time(&msg_ptr->end_time, buffer);
safe_unpack32(&msg_ptr->exit_code, buffer);
safe_unpack32(&msg_ptr->job_id, buffer);
safe_unpack32(&msg_ptr->job_state, buffer);
safe_unpackstr_xmalloc(&msg_ptr->nodes, &uint32_tmp, buffer);
safe_unpack32(&msg_ptr->req_uid, buffer);
safe_unpack_time(&msg_ptr->start_time, buffer);
safe_unpack_time(&msg_ptr->submit_time, buffer);
safe_unpackstr_xmalloc(&msg_ptr->system_comment,
&uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_alloc_str,
&uint32_tmp, buffer);
} else
goto unpack_error;
return SLURM_SUCCESS;
unpack_error:
slurmdbd_free_job_complete_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
static int _unpack_step_complete_msg(dbd_step_comp_msg_t **msg,
uint16_t rpc_version, Buf buffer)
{
uint32_t uint32_tmp;
dbd_step_comp_msg_t *msg_ptr = xmalloc(sizeof(dbd_step_comp_msg_t));
*msg = msg_ptr;
if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpack32(&msg_ptr->assoc_id, buffer);
safe_unpack64(&msg_ptr->db_index, buffer);
safe_unpack_time(&msg_ptr->end_time, buffer);
safe_unpack32(&msg_ptr->exit_code, buffer);
jobacctinfo_unpack((struct jobacctinfo **)&msg_ptr->jobacct,
rpc_version, PROTOCOL_TYPE_DBD, buffer, 1);
safe_unpack32(&msg_ptr->job_id, buffer);
safe_unpack_time(&msg_ptr->job_submit_time, buffer);
safe_unpackstr_xmalloc(&msg_ptr->job_tres_alloc_str,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->req_uid, buffer);
safe_unpack_time(&msg_ptr->start_time, buffer);
safe_unpack16(&msg_ptr->state, buffer);
safe_unpack32(&msg_ptr->step_id, buffer);
safe_unpack32(&msg_ptr->total_tasks, buffer);
} else
goto unpack_error;
return SLURM_SUCCESS;
unpack_error:
debug2("slurmdbd_unpack_step_complete_msg:"
"unpack_error: size_buf(buffer) %u",
size_buf(buffer));
slurmdbd_free_step_complete_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
extern int ext_sensors_data_unpack(ext_sensors_data_t **ext_sensors, Buf buffer,
uint16_t protocol_version)
{
ext_sensors_data_t *ext_sensors_ptr = ext_sensors_alloc();
*ext_sensors = ext_sensors_ptr;
if (ext_sensors_ptr == NULL)
return SLURM_ERROR;
safe_unpack32(&ext_sensors_ptr->consumed_energy, buffer);
safe_unpack32(&ext_sensors_ptr->temperature, buffer);
safe_unpack_time(&ext_sensors_ptr->energy_update_time, buffer);
safe_unpack32(&ext_sensors_ptr->current_watts, buffer);
return SLURM_SUCCESS;
unpack_error:
ext_sensors_destroy(ext_sensors_ptr);
*ext_sensors = NULL;
return SLURM_ERROR;
}
static int _unpack_cluster_tres_msg(dbd_cluster_tres_msg_t **msg,
uint16_t rpc_version, Buf buffer)
{
dbd_cluster_tres_msg_t *msg_ptr;
uint32_t uint32_tmp;
msg_ptr = xmalloc(sizeof(dbd_cluster_tres_msg_t));
*msg = msg_ptr;
if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
safe_unpackstr_xmalloc(&msg_ptr->cluster_nodes,
&uint32_tmp, buffer);
safe_unpack_time(&msg_ptr->event_time, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_str,
&uint32_tmp, buffer);
}
return SLURM_SUCCESS;
unpack_error:
slurmdbd_free_cluster_tres_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
/*
* load_all_part_state - load the partition state from file, recover on
* slurmctld restart. execute this after loading the configuration
* file data.
* NOTE: READ lock_slurmctld config before entry
*/
int load_all_part_state(void)
{
char *part_name = NULL, *allow_groups = NULL, *nodes = NULL;
char *state_file, *data = NULL;
uint32_t max_time, default_time, max_nodes, min_nodes;
uint32_t max_cpus_per_node = INFINITE, grace_time = 0;
time_t time;
uint16_t flags;
uint16_t max_share, preempt_mode, priority, state_up, cr_type;
struct part_record *part_ptr;
uint32_t data_size = 0, name_len;
int data_allocated, data_read = 0, error_code = 0, part_cnt = 0;
int state_fd;
Buf buffer;
char *ver_str = NULL;
char* allow_alloc_nodes = NULL;
uint16_t protocol_version = (uint16_t)NO_VAL;
char* alternate = NULL;
/* read the file */
lock_state_files();
state_fd = _open_part_state_file(&state_file);
if (state_fd < 0) {
info("No partition state file (%s) to recover",
state_file);
error_code = ENOENT;
} else {
data_allocated = BUF_SIZE;
data = xmalloc(data_allocated);
while (1) {
data_read = read(state_fd, &data[data_size],
BUF_SIZE);
if (data_read < 0) {
if (errno == EINTR)
continue;
else {
error("Read error on %s: %m",
state_file);
break;
}
} else if (data_read == 0) /* eof */
break;
data_size += data_read;
data_allocated += data_read;
xrealloc(data, data_allocated);
}
close(state_fd);
}
xfree(state_file);
unlock_state_files();
buffer = create_buf(data, data_size);
safe_unpackstr_xmalloc( &ver_str, &name_len, buffer);
debug3("Version string in part_state header is %s", ver_str);
if (ver_str) {
if (!strcmp(ver_str, PART_STATE_VERSION)) {
protocol_version = SLURM_PROTOCOL_VERSION;
} else if (!strcmp(ver_str, PART_2_5_STATE_VERSION)) {
protocol_version = SLURM_2_5_PROTOCOL_VERSION;
}
}
if (protocol_version == (uint16_t)NO_VAL) {
error("**********************************************************");
error("Can not recover partition state, data version incompatible");
error("**********************************************************");
xfree(ver_str);
free_buf(buffer);
return EFAULT;
}
xfree(ver_str);
safe_unpack_time(&time, buffer);
while (remaining_buf(buffer) > 0) {
if (protocol_version >= SLURM_2_6_PROTOCOL_VERSION) {
safe_unpackstr_xmalloc(&part_name, &name_len, buffer);
safe_unpack32(&grace_time, buffer);
safe_unpack32(&max_time, buffer);
safe_unpack32(&default_time, buffer);
safe_unpack32(&max_cpus_per_node, buffer);
safe_unpack32(&max_nodes, buffer);
safe_unpack32(&min_nodes, buffer);
safe_unpack16(&flags, buffer);
safe_unpack16(&max_share, buffer);
safe_unpack16(&preempt_mode, buffer);
safe_unpack16(&priority, buffer);
if (priority > part_max_priority)
part_max_priority = priority;
safe_unpack16(&state_up, buffer);
safe_unpack16(&cr_type, buffer);
safe_unpackstr_xmalloc(&allow_groups,
&name_len, buffer);
safe_unpackstr_xmalloc(&allow_alloc_nodes,
&name_len, buffer);
safe_unpackstr_xmalloc(&alternate, &name_len, buffer);
safe_unpackstr_xmalloc(&nodes, &name_len, buffer);
if ((flags & PART_FLAG_DEFAULT_CLR) ||
(flags & PART_FLAG_HIDDEN_CLR) ||
(flags & PART_FLAG_NO_ROOT_CLR) ||
(flags & PART_FLAG_ROOT_ONLY_CLR) ||
(flags & PART_FLAG_REQ_RESV_CLR)) {
error("Invalid data for partition %s: flags=%u",
part_name, flags);
error_code = EINVAL;
}
} else if (protocol_version >= SLURM_2_4_PROTOCOL_VERSION) {
safe_unpackstr_xmalloc(&part_name, &name_len, buffer);
safe_unpack32(&grace_time, buffer);
safe_unpack32(&max_time, buffer);
safe_unpack32(&default_time, buffer);
safe_unpack32(&max_nodes, buffer);
safe_unpack32(&min_nodes, buffer);
safe_unpack16(&flags, buffer);
safe_unpack16(&max_share, buffer);
safe_unpack16(&preempt_mode, buffer);
safe_unpack16(&priority, buffer);
if (priority > part_max_priority)
part_max_priority = priority;
cr_type = 0; /* Default value */
safe_unpack16(&state_up, buffer);
safe_unpackstr_xmalloc(&allow_groups,
&name_len, buffer);
safe_unpackstr_xmalloc(&allow_alloc_nodes,
&name_len, buffer);
safe_unpackstr_xmalloc(&alternate, &name_len, buffer);
safe_unpackstr_xmalloc(&nodes, &name_len, buffer);
if ((flags & PART_FLAG_DEFAULT_CLR) ||
(flags & PART_FLAG_HIDDEN_CLR) ||
(flags & PART_FLAG_NO_ROOT_CLR) ||
(flags & PART_FLAG_ROOT_ONLY_CLR) ||
(flags & PART_FLAG_REQ_RESV_CLR)) {
error("Invalid data for partition %s: flags=%u",
part_name, flags);
error_code = EINVAL;
}
} else {
error("load_all_part_state: protocol_version "
"%hu not supported", protocol_version);
goto unpack_error;
}
/* validity test as possible */
if (state_up > PARTITION_UP) {
error("Invalid data for partition %s: state_up=%u",
part_name, state_up);
error_code = EINVAL;
}
if (error_code) {
error("No more partition data will be processed from "
"the checkpoint file");
xfree(allow_groups);
xfree(allow_alloc_nodes);
xfree(alternate);
xfree(part_name);
xfree(nodes);
error_code = EINVAL;
break;
}
/* find record and perform update */
part_ptr = list_find_first(part_list, &list_find_part,
part_name);
part_cnt++;
if (part_ptr == NULL) {
info("load_all_part_state: partition %s missing from "
"configuration file", part_name);
part_ptr = create_part_record();
xfree(part_ptr->name);
part_ptr->name = xstrdup(part_name);
}
part_ptr->flags = flags;
if (part_ptr->flags & PART_FLAG_DEFAULT) {
xfree(default_part_name);
default_part_name = xstrdup(part_name);
default_part_loc = part_ptr;
}
part_ptr->max_time = max_time;
part_ptr->default_time = default_time;
part_ptr->max_cpus_per_node = max_cpus_per_node;
part_ptr->max_nodes = max_nodes;
part_ptr->max_nodes_orig = max_nodes;
part_ptr->min_nodes = min_nodes;
part_ptr->min_nodes_orig = min_nodes;
part_ptr->max_share = max_share;
part_ptr->grace_time = grace_time;
if (preempt_mode != (uint16_t) NO_VAL)
part_ptr->preempt_mode = preempt_mode;
part_ptr->priority = priority;
part_ptr->state_up = state_up;
part_ptr->cr_type = cr_type;
xfree(part_ptr->allow_groups);
part_ptr->allow_groups = allow_groups;
xfree(part_ptr->allow_alloc_nodes);
part_ptr->allow_alloc_nodes = allow_alloc_nodes;
xfree(part_ptr->alternate);
part_ptr->alternate = alternate;
xfree(part_ptr->nodes);
part_ptr->nodes = nodes;
xfree(part_name);
}
info("Recovered state of %d partitions", part_cnt);
free_buf(buffer);
return error_code;
unpack_error:
error("Incomplete partition data checkpoint file");
info("Recovered state of %d partitions", part_cnt);
free_buf(buffer);
return EFAULT;
}
/*
* load_all_front_end_state - Load the front_end node state from file, recover
* on slurmctld restart. Execute this after loading the configuration
* file data. Data goes into common storage.
* IN state_only - if true, overwrite only front_end node state and reason
* Use this to overwrite the "UNKNOWN state typically used in slurm.conf
* RET 0 or error code
* NOTE: READ lock_slurmctld config before entry
*/
extern int load_all_front_end_state(bool state_only)
{
#ifdef HAVE_FRONT_END
char *node_name = NULL, *reason = NULL, *data = NULL, *state_file;
int data_allocated, data_read = 0, error_code = 0, node_cnt = 0;
uint16_t node_state;
uint32_t data_size = 0, name_len;
uint32_t reason_uid = NO_VAL;
time_t reason_time = 0;
front_end_record_t *front_end_ptr;
int state_fd;
time_t time_stamp;
Buf buffer;
char *ver_str = NULL;
uint16_t protocol_version = (uint16_t) NO_VAL;
/* read the file */
lock_state_files ();
state_fd = _open_front_end_state_file(&state_file);
if (state_fd < 0) {
info ("No node state file (%s) to recover", state_file);
error_code = ENOENT;
} else {
data_allocated = BUF_SIZE;
data = xmalloc(data_allocated);
while (1) {
data_read = read(state_fd, &data[data_size], BUF_SIZE);
if (data_read < 0) {
if (errno == EINTR)
continue;
else {
error ("Read error on %s: %m",
state_file);
break;
}
} else if (data_read == 0) /* eof */
break;
data_size += data_read;
data_allocated += data_read;
xrealloc(data, data_allocated);
}
close (state_fd);
}
xfree (state_file);
unlock_state_files ();
buffer = create_buf (data, data_size);
safe_unpackstr_xmalloc( &ver_str, &name_len, buffer);
debug3("Version string in front_end_state header is %s", ver_str);
if (ver_str) {
if (!strcmp(ver_str, FRONT_END_STATE_VERSION)) {
protocol_version = SLURM_PROTOCOL_VERSION;
}
}
if (protocol_version == (uint16_t) NO_VAL) {
error("*****************************************************");
error("Can not recover front_end state, version incompatible");
error("*****************************************************");
xfree(ver_str);
free_buf(buffer);
return EFAULT;
}
xfree(ver_str);
safe_unpack_time(&time_stamp, buffer);
while (remaining_buf (buffer) > 0) {
uint16_t base_state;
if (protocol_version >= SLURM_2_5_PROTOCOL_VERSION) {
safe_unpackstr_xmalloc (&node_name, &name_len, buffer);
safe_unpack16 (&node_state, buffer);
safe_unpackstr_xmalloc (&reason, &name_len, buffer);
safe_unpack_time (&reason_time, buffer);
safe_unpack32 (&reason_uid, buffer);
base_state = node_state & NODE_STATE_BASE;
} else
goto unpack_error;
/* validity test as possible */
/* find record and perform update */
front_end_ptr = find_front_end_record(node_name);
if (front_end_ptr == NULL) {
error("Front_end node %s has vanished from "
"configuration", node_name);
} else if (state_only) {
uint16_t orig_flags;
orig_flags = front_end_ptr->node_state &
NODE_STATE_FLAGS;
node_cnt++;
if (IS_NODE_UNKNOWN(front_end_ptr)) {
if (base_state == NODE_STATE_DOWN) {
orig_flags &= (~NODE_STATE_COMPLETING);
front_end_ptr->node_state =
NODE_STATE_DOWN | orig_flags;
}
if (node_state & NODE_STATE_DRAIN) {
front_end_ptr->node_state |=
NODE_STATE_DRAIN;
}
if (node_state & NODE_STATE_FAIL) {
front_end_ptr->node_state |=
NODE_STATE_FAIL;
}
}
if (front_end_ptr->reason == NULL) {
front_end_ptr->reason = reason;
reason = NULL; /* Nothing to free */
front_end_ptr->reason_time = reason_time;
front_end_ptr->reason_uid = reason_uid;
}
} else {
node_cnt++;
front_end_ptr->node_state = node_state;
xfree(front_end_ptr->reason);
front_end_ptr->reason = reason;
reason = NULL; /* Nothing to free */
front_end_ptr->reason_time = reason_time;
front_end_ptr->reason_uid = reason_uid;
front_end_ptr->last_response = (time_t) 0;
}
xfree(node_name);
xfree(reason);
}
fini: info("Recovered state of %d front_end nodes", node_cnt);
free_buf (buffer);
return error_code;
unpack_error:
error("Incomplete front_end node data checkpoint file");
error_code = EFAULT;
xfree (node_name);
xfree(reason);
goto fini;
#else
return 0;
#endif
}
extern slurmdb_federation_rec_t *fed_mgr_state_load(char *state_save_location)
{
Buf buffer = NULL;
char *data = NULL, *state_file;
time_t buf_time;
uint16_t ver = 0;
uint32_t data_size = 0;
int state_fd;
int data_allocated, data_read = 0, error_code = SLURM_SUCCESS;
slurmdb_federation_rec_t *ret_fed = NULL;
state_file = xstrdup_printf("%s/%s", state_save_location,
FED_MGR_STATE_FILE);
state_fd = open(state_file, O_RDONLY);
if (state_fd < 0) {
error("No fed_mgr state file (%s) to recover", state_file);
xfree(state_file);
return NULL;
} else {
data_allocated = BUF_SIZE;
data = xmalloc(data_allocated);
while (1) {
data_read = read(state_fd, &data[data_size],
BUF_SIZE);
if (data_read < 0) {
if (errno == EINTR)
continue;
else {
error("Read error on %s: %m",
state_file);
break;
}
} else if (data_read == 0) /* eof */
break;
data_size += data_read;
data_allocated += data_read;
xrealloc(data, data_allocated);
}
close(state_fd);
}
xfree(state_file);
buffer = create_buf(data, data_size);
safe_unpack16(&ver, buffer);
debug3("Version in fed_mgr_state header is %u", ver);
if (ver > SLURM_PROTOCOL_VERSION || ver < SLURM_MIN_PROTOCOL_VERSION) {
error("***********************************************");
error("Can not recover fed_mgr state, incompatible version, "
"got %u need > %u <= %u", ver,
SLURM_MIN_PROTOCOL_VERSION, SLURM_PROTOCOL_VERSION);
error("***********************************************");
free_buf(buffer);
return NULL;
}
safe_unpack_time(&buf_time, buffer);
error_code = slurmdb_unpack_federation_rec((void **)&ret_fed, ver,
buffer);
if (error_code != SLURM_SUCCESS)
goto unpack_error;
else if (!ret_fed || !ret_fed->name ||
!list_count(ret_fed->cluster_list)) {
slurmdb_destroy_federation_rec(ret_fed);
ret_fed = NULL;
error("No feds retrieved");
} else {
/* We want to free the connections here since they don't exist
* anymore, but they were packed when state was saved. */
slurmdb_cluster_rec_t *cluster;
ListIterator itr = list_iterator_create(
ret_fed->cluster_list);
while ((cluster = list_next(itr))) {
slurm_persist_conn_destroy(cluster->fed.recv);
cluster->fed.recv = NULL;
slurm_persist_conn_destroy(cluster->fed.send);
cluster->fed.send = NULL;
}
list_iterator_destroy(itr);
}
free_buf(buffer);
return ret_fed;
unpack_error:
free_buf(buffer);
return NULL;
}
static void _read_last_decay_ran(time_t *last_ran, time_t *last_reset)
{
int data_allocated, data_read = 0;
uint32_t data_size = 0;
int state_fd;
char *data = NULL, *state_file;
Buf buffer;
xassert(last_ran);
xassert(last_reset);
(*last_ran) = 0;
(*last_reset) = 0;
/* read the file */
state_file = xstrdup(slurmctld_conf.state_save_location);
xstrcat(state_file, "/priority_last_decay_ran");
lock_state_files();
state_fd = open(state_file, O_RDONLY);
if (state_fd < 0) {
info("No last decay (%s) to recover", state_file);
unlock_state_files();
return;
} else {
data_allocated = BUF_SIZE;
data = xmalloc(data_allocated);
while (1) {
data_read = read(state_fd, &data[data_size],
BUF_SIZE);
if (data_read < 0) {
if (errno == EINTR)
continue;
else {
error("Read error on %s: %m",
state_file);
break;
}
} else if (data_read == 0) /* eof */
break;
data_size += data_read;
data_allocated += data_read;
xrealloc(data, data_allocated);
}
close(state_fd);
}
xfree(state_file);
unlock_state_files();
buffer = create_buf(data, data_size);
safe_unpack_time(last_ran, buffer);
safe_unpack_time(last_reset, buffer);
free_buf(buffer);
if (priority_debug)
info("Last ran decay on jobs at %ld", (long)*last_ran);
return;
unpack_error:
error("Incomplete priority last decay file returning");
free_buf(buffer);
return;
}
static void _load_sicp_state(void)
{
int data_allocated, data_read = 0;
uint32_t data_size = 0;
int state_fd, sicp_cnt = 0;
char *data = NULL, *state_file;
struct stat stat_buf;
Buf buffer;
char *ver_str = NULL;
uint32_t ver_str_len;
uint16_t protocol_version = (uint16_t)NO_VAL;
uint32_t job_id = 0;
uint32_t job_state = 0;
sicp_job_t *sicp_ptr;
time_t buf_time, now;
/* read the file */
lock_state_files();
state_file = xstrdup(slurmctld_conf.state_save_location);
xstrcat(state_file, "/sicp_state");
state_fd = open(state_file, O_RDONLY);
if (state_fd < 0) {
error("Could not open job state file %s: %m", state_file);
unlock_state_files();
xfree(state_file);
return;
} else if (fstat(state_fd, &stat_buf) < 0) {
error("Could not stat job state file %s: %m", state_file);
unlock_state_files();
(void) close(state_fd);
xfree(state_file);
return;
} else if (stat_buf.st_size < 10) {
error("Job state file %s too small", state_file);
unlock_state_files();
(void) close(state_fd);
xfree(state_file);
return;
}
data_allocated = BUF_SIZE;
data = xmalloc(data_allocated);
while (1) {
data_read = read(state_fd, &data[data_size], BUF_SIZE);
if (data_read < 0) {
if (errno == EINTR)
continue;
else {
error("Read error on %s: %m", state_file);
break;
}
} else if (data_read == 0) /* eof */
break;
data_size += data_read;
data_allocated += data_read;
xrealloc(data, data_allocated);
}
close(state_fd);
xfree(state_file);
unlock_state_files();
buffer = create_buf(data, data_size);
safe_unpackstr_xmalloc(&ver_str, &ver_str_len, buffer);
debug3("Version string in sicp_state header is %s", ver_str);
if (ver_str && !strcmp(ver_str, "PROTOCOL_VERSION"))
safe_unpack16(&protocol_version, buffer);
xfree(ver_str);
if (protocol_version == (uint16_t)NO_VAL) {
error("************************************************");
error("Can not recover SICP state, incompatible version");
error("************************************************");
xfree(ver_str);
free_buf(buffer);
return;
}
safe_unpack_time(&buf_time, buffer);
now = time(NULL);
while (remaining_buf(buffer) > 0) {
safe_unpack32(&job_id, buffer);
safe_unpack32(&job_state, buffer);
sicp_ptr = xmalloc(sizeof(sicp_job_t));
sicp_ptr->job_id = job_id;
sicp_ptr->job_state = job_state;
sicp_ptr->update_time = now;
list_append(sicp_job_list, sicp_ptr);
_add_job_hash(sicp_ptr);
sicp_cnt++;
}
free_buf(buffer);
info("Recovered information about %d sicp jobs", sicp_cnt);
if (slurm_get_debug_flags() & DEBUG_FLAG_SICP)
_log_sicp_recs();
return;
unpack_error:
error("Incomplete sicp data checkpoint file");
info("Recovered information about %d sicp jobs", sicp_cnt);
free_buf(buffer);
return;
}
static int _unpack_job_start_msg(void **msg,
uint16_t rpc_version, Buf buffer)
{
uint32_t uint32_tmp;
dbd_job_start_msg_t *msg_ptr = xmalloc(sizeof(dbd_job_start_msg_t));
*msg = msg_ptr;
msg_ptr->array_job_id = 0;
msg_ptr->array_task_id = NO_VAL;
if (rpc_version >= SLURM_19_05_PROTOCOL_VERSION) {
safe_unpackstr_xmalloc(&msg_ptr->account, &uint32_tmp, buffer);
safe_unpack32(&msg_ptr->alloc_nodes, buffer);
safe_unpack32(&msg_ptr->array_job_id, buffer);
safe_unpack32(&msg_ptr->array_max_tasks, buffer);
safe_unpack32(&msg_ptr->array_task_id, buffer);
safe_unpackstr_xmalloc(&msg_ptr->array_task_str,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->array_task_pending, buffer);
safe_unpack32(&msg_ptr->assoc_id, buffer);
safe_unpackstr_xmalloc(&msg_ptr->constraints,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->db_flags, buffer);
safe_unpack64(&msg_ptr->db_index, buffer);
safe_unpack_time(&msg_ptr->eligible_time, buffer);
safe_unpack32(&msg_ptr->gid, buffer);
safe_unpackstr_xmalloc(&msg_ptr->gres_alloc, &uint32_tmp,
buffer);
safe_unpackstr_xmalloc(&msg_ptr->gres_req, &uint32_tmp,
buffer);
safe_unpackstr_xmalloc(&msg_ptr->gres_used, &uint32_tmp,
buffer);
safe_unpack32(&msg_ptr->job_id, buffer);
safe_unpack32(&msg_ptr->job_state, buffer);
safe_unpack32(&msg_ptr->state_reason_prev, buffer);
safe_unpackstr_xmalloc(&msg_ptr->mcs_label,
&uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->name, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->nodes, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->node_inx, &uint32_tmp, buffer);
safe_unpack32(&msg_ptr->pack_job_id, buffer);
safe_unpack32(&msg_ptr->pack_job_offset, buffer);
safe_unpackstr_xmalloc(&msg_ptr->partition,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->priority, buffer);
safe_unpack32(&msg_ptr->qos_id, buffer);
safe_unpack32(&msg_ptr->req_cpus, buffer);
safe_unpack64(&msg_ptr->req_mem, buffer);
safe_unpack32(&msg_ptr->resv_id, buffer);
safe_unpack_time(&msg_ptr->start_time, buffer);
safe_unpack_time(&msg_ptr->submit_time, buffer);
safe_unpack32(&msg_ptr->timelimit, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_alloc_str,
&uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_req_str,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->uid, buffer);
safe_unpackstr_xmalloc(&msg_ptr->wckey, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->work_dir, &uint32_tmp, buffer);
} else if (rpc_version >= SLURM_MIN_PROTOCOL_VERSION) {
char *tmp_char = NULL;
safe_unpackstr_xmalloc(&msg_ptr->account, &uint32_tmp, buffer);
safe_unpack32(&msg_ptr->alloc_nodes, buffer);
safe_unpack32(&msg_ptr->array_job_id, buffer);
safe_unpack32(&msg_ptr->array_max_tasks, buffer);
safe_unpack32(&msg_ptr->array_task_id, buffer);
safe_unpackstr_xmalloc(&msg_ptr->array_task_str,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->array_task_pending, buffer);
safe_unpack32(&msg_ptr->assoc_id, buffer);
safe_unpackstr_xmalloc(&tmp_char, &uint32_tmp, buffer);
xfree(tmp_char); /* block_id */
safe_unpack64(&msg_ptr->db_index, buffer);
safe_unpack_time(&msg_ptr->eligible_time, buffer);
safe_unpack32(&msg_ptr->gid, buffer);
safe_unpackstr_xmalloc(&msg_ptr->gres_alloc, &uint32_tmp,
buffer);
safe_unpackstr_xmalloc(&msg_ptr->gres_req, &uint32_tmp,
buffer);
safe_unpackstr_xmalloc(&msg_ptr->gres_used, &uint32_tmp,
buffer);
safe_unpack32(&msg_ptr->job_id, buffer);
safe_unpack32(&msg_ptr->job_state, buffer);
safe_unpackstr_xmalloc(&msg_ptr->mcs_label,
&uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->name, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->nodes, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->node_inx, &uint32_tmp, buffer);
safe_unpack32(&msg_ptr->pack_job_id, buffer);
safe_unpack32(&msg_ptr->pack_job_offset, buffer);
safe_unpackstr_xmalloc(&msg_ptr->partition,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->priority, buffer);
safe_unpack32(&msg_ptr->qos_id, buffer);
safe_unpack32(&msg_ptr->req_cpus, buffer);
safe_unpack64(&msg_ptr->req_mem, buffer);
safe_unpack32(&msg_ptr->resv_id, buffer);
safe_unpack_time(&msg_ptr->start_time, buffer);
safe_unpack_time(&msg_ptr->submit_time, buffer);
safe_unpack32(&msg_ptr->timelimit, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_alloc_str,
&uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->tres_req_str,
&uint32_tmp, buffer);
safe_unpack32(&msg_ptr->uid, buffer);
safe_unpackstr_xmalloc(&msg_ptr->wckey, &uint32_tmp, buffer);
safe_unpackstr_xmalloc(&msg_ptr->work_dir, &uint32_tmp, buffer);
} else
goto unpack_error;
return SLURM_SUCCESS;
unpack_error:
slurmdbd_free_job_start_msg(msg_ptr);
*msg = NULL;
return SLURM_ERROR;
}
