dailz/hpc
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(service): map additional Slurm SDK fields and fix ExitCode/Default bugs

Browse Source 
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
This commit is contained in:
dailz
2026-04-10 11:12:51 +08:00
parent 85901fe18a
commit 824d9e816f
2 changed files with 146 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

91
internal/service/cluster_service.go
View File

@@ -46,6 +46,27 @@ func uint32NoValString(v *slurm.Uint32NoVal) string {
return ""
}
func derefUint64NoValInt64(v *slurm.Uint64NoVal) *int64 {
if v != nil && v.Number != nil {
return v.Number
}
return nil
}
func derefCSVString(cs *slurm.CSVString) string {
if cs == nil || len(*cs) == 0 {
return ""
}
result := ""
for i, s := range *cs {
if i > 0 {
result += ","
}
result += s
}
return result
}
type ClusterService struct {
client *slurm.Client
logger *zap.Logger
@@ -230,17 +251,39 @@ func mapNode(n slurm.Node) model.NodeResponse {
Name: derefStr(n.Name),
State: n.State,
CPUs: derefInt32(n.Cpus),
AllocCpus: n.AllocCpus,
Cores: n.Cores,
Sockets: n.Sockets,
Threads: n.Threads,
RealMemory: derefInt64(n.RealMemory),
AllocMem: derefInt64(n.AllocMemory),
AllocMemory: derefInt64(n.AllocMemory),
FreeMem: derefUint64NoValInt64(n.FreeMem),
CpuLoad: n.CpuLoad,
Arch: derefStr(n.Architecture),
OS: derefStr(n.OperatingSystem),
Gres: derefStr(n.Gres),
GresUsed: derefStr(n.GresUsed),
Reason: derefStr(n.Reason),
ReasonSetByUser: derefStr(n.ReasonSetByUser),
Address: derefStr(n.Address),
Hostname: derefStr(n.Hostname),
Weight: n.Weight,
Features: derefCSVString(n.Features),
ActiveFeatures: derefCSVString(n.ActiveFeatures),
}
}
func mapPartition(pi slurm.PartitionInfo) model.PartitionResponse {
var state []string
var isDefault bool
if pi.Partition != nil {
state = pi.Partition.State
for _, s := range state {
if s == "DEFAULT" {
isDefault = true
break
}
}
}
var nodes string
if pi.Nodes != nil {
@@ -258,12 +301,56 @@ func mapPartition(pi slurm.PartitionInfo) model.PartitionResponse {
if pi.Maximums != nil {
maxTime = uint32NoValString(pi.Maximums.Time)
}
var maxNodes *int32
if pi.Maximums != nil {
maxNodes = mapUint32NoValToInt32(pi.Maximums.Nodes)
}
var maxCPUsPerNode *int32
if pi.Maximums != nil {
maxCPUsPerNode = pi.Maximums.CpusPerNode
}
var minNodes *int32
if pi.Minimums != nil {
minNodes = pi.Minimums.Nodes
}
var defaultTime string
if pi.Defaults != nil {
defaultTime = uint32NoValString(pi.Defaults.Time)
}
var graceTime *int32 = pi.GraceTime
var priority *int32
if pi.Priority != nil {
priority = pi.Priority.JobFactor
}
var qosAllowed, qosDeny, qosAssigned string
if pi.QOS != nil {
qosAllowed = derefStr(pi.QOS.Allowed)
qosDeny = derefStr(pi.QOS.Deny)
qosAssigned = derefStr(pi.QOS.Assigned)
}
var accountsAllowed, accountsDeny string
if pi.Accounts != nil {
accountsAllowed = derefStr(pi.Accounts.Allowed)
accountsDeny = derefStr(pi.Accounts.Deny)
}
return model.PartitionResponse{
Name: derefStr(pi.Name),
State: state,
Default: isDefault,
Nodes: nodes,
TotalCPUs: totalCPUs,
TotalNodes: totalNodes,
TotalCPUs: totalCPUs,
MaxTime: maxTime,
MaxNodes: maxNodes,
MaxCPUsPerNode: maxCPUsPerNode,
MinNodes: minNodes,
DefaultTime: defaultTime,
GraceTime: graceTime,
Priority: priority,
QOSAllowed: qosAllowed,
QOSDeny: qosDeny,
QOSAssigned: qosAssigned,
AccountsAllowed: accountsAllowed,
AccountsDeny: accountsDeny,
}
}

46
internal/service/job_service.go
View File

@@ -282,6 +282,14 @@ func strToPtrOrNil(s string) *string {
return &s
}
func mapUint32NoValToInt32(v *slurm.Uint32NoVal) *int32 {
if v != nil && v.Number != nil {
n := int32(*v.Number)
return &n
}
return nil
}
// mapJobInfo maps SDK JobInfo to API JobResponse.
func mapJobInfo(ji *slurm.JobInfo) model.JobResponse {
resp := model.JobResponse{}
@@ -295,6 +303,17 @@ func mapJobInfo(ji *slurm.JobInfo) model.JobResponse {
if ji.Partition != nil {
resp.Partition = *ji.Partition
}
resp.Account = derefStr(ji.Account)
resp.User = derefStr(ji.UserName)
resp.Cluster = derefStr(ji.Cluster)
resp.QOS = derefStr(ji.Qos)
resp.Priority = mapUint32NoValToInt32(ji.Priority)
resp.TimeLimit = uint32NoValString(ji.TimeLimit)
resp.StateReason = derefStr(ji.StateReason)
resp.Cpus = mapUint32NoValToInt32(ji.Cpus)
resp.Tasks = mapUint32NoValToInt32(ji.Tasks)
resp.NodeCount = mapUint32NoValToInt32(ji.NodeCount)
resp.BatchHost = derefStr(ji.BatchHost)
if ji.SubmitTime != nil && ji.SubmitTime.Number != nil {
resp.SubmitTime = ji.SubmitTime.Number
}
@@ -311,6 +330,13 @@ func mapJobInfo(ji *slurm.JobInfo) model.JobResponse {
if ji.Nodes != nil {
resp.Nodes = *ji.Nodes
}
resp.StdOut = derefStr(ji.StandardOutput)
resp.StdErr = derefStr(ji.StandardError)
resp.StdIn = derefStr(ji.StandardInput)
resp.WorkDir = derefStr(ji.CurrentWorkingDirectory)
resp.Command = derefStr(ji.Command)
resp.ArrayJobID = mapUint32NoValToInt32(ji.ArrayJobID)
resp.ArrayTaskID = mapUint32NoValToInt32(ji.ArrayTaskID)
return resp
}
@@ -325,11 +351,20 @@ func mapSlurmdbJob(j *slurm.Job) model.JobResponse {
}
if j.State != nil {
resp.State = j.State.Current
resp.StateReason = derefStr(j.State.Reason)
}
if j.Partition != nil {
resp.Partition = *j.Partition
}
resp.Account = derefStr(j.Account)
if j.User != nil {
resp.User = *j.User
}
resp.Cluster = derefStr(j.Cluster)
resp.QOS = derefStr(j.Qos)
resp.Priority = mapUint32NoValToInt32(j.Priority)
if j.Time != nil {
resp.TimeLimit = uint32NoValString(j.Time.Limit)
if j.Time.Submission != nil {
resp.SubmitTime = j.Time.Submission
}
@@ -340,8 +375,19 @@ func mapSlurmdbJob(j *slurm.Job) model.JobResponse {
resp.EndTime = j.Time.End
}
}
if j.ExitCode != nil && j.ExitCode.ReturnCode != nil && j.ExitCode.ReturnCode.Number != nil {
code := int32(*j.ExitCode.ReturnCode.Number)
resp.ExitCode = &code
}
if j.Nodes != nil {
resp.Nodes = *j.Nodes
}
if j.Required != nil {
resp.Cpus = j.Required.CPUs
}
if j.AllocationNodes != nil {
resp.NodeCount = j.AllocationNodes
}
resp.WorkDir = derefStr(j.WorkingDirectory)
return resp
}