- slurm-users - lists.schedmd.com

controller backup slurmctld error while takeover
by Miriam Olmi 26 Mar '24

26 Mar '24

Dear all, I am having trouble finalizing the configuration of the backup controller for my slurm cluster. In principle, if no job is running everything seems fine: both the slurmctld services on the primary and the backup controller are running and if I stop the service on the primary controller after 10s more or less (SlurmctldTimeout = 10 sec) the backup controller takes over. Also, if I run the sinfo or squeue command during the 10s of inactivity, the shell stay pending but it recover perfectly after the time needed by the backup controller to take control and it works the same when the primary controller is back. Unfortunately, if I try to do the same test while a job is running there are two different behaviors depending on the initial scenario. 1st scenario: Both the primary and backup controller are fine. I launch a batch script and I verify the script is running with sinfo and squeue. While the script is still running I stop the service on the primary controller with success but at this point everything gets crazy: on the backup controller in the slurmctld service log I find the following errors: slurmctld: error: Invalid RPC received REQUEST_JOB_INFO while in standby mode slurmctld: error: Invalid RPC received REQUEST_PARTITION_INFO while in standby mode slurmctld: error: Invalid RPC received REQUEST_JOB_INFO while in standby mode slurmctld: error: Invalid RPC received REQUEST_PARTITION_INFO while in standby mode slurmctld: error: slurm_accept_msg_conn poll: Bad address slurmctld: error: slurm_accept_msg_conn poll: Bad address and the commands sinfo and squeue are Unable to contact slurm controller (connect failure). 2nd scenario: the primary controller is stopped and I launch a batch job while the backup controller is the only one working. While the job is running, I restart the slurmctld service on the primary controller. In this case the primary controller takes over immediately: everything is smooth and safe and the sinfo and squeue commands continue to work perfectly. What might be the problem? Many thanks in advance! Miriam

2 5

cgroups_exporter for slurm on rhel9 (cgroups-v2)
by Saluja, Prabhjyot 25 Mar '24

25 Mar '24

Hi All, We are currently trying to set up cgroup_exporter <https://github.com/treydock/cgroup_exporter> for slurm. It's been working smoothly with cgroups.v1 and slurm-22.05.7. However, we're facing some challenges with RHEL-9, slurm-23.11.1 and cgroups.v2. The cgroup_exporter isn't capturing the slurm cgroup job information. I'm reaching out to see if any other sites have managed to make this work. If you're using a different exporter that's working for your site, could you please let us know? Thanks! Regards, Singh

2 1

Slurm version 23.11.5 is now available
by Tim McMullan 22 Mar '24

22 Mar '24

We are pleased to announce the availability of Slurm version 23.11.5. The 23.11.5 release includes some important fixes related to newer features as well as some database fixes. The most noteworthy fixes include fixing the sattach command (which only worked for root and SlurmUser after 23.11.0) and fixing an issue while constructing the new lineage database entries. This last change will also perform a query during the upgrade from any prior 23.11 version to fix existing databases. Slurm can be downloaded from https://www.schedmd.com/downloads.php . -Tim > * Changes in Slurm 23.11.5 > ========================== > -- Fix Debian package build on systems that are not able to query the systemd > package. > -- data_parser/v0.0.40 - Emit a warning instead of an error if a disabled > parser is invoked. > -- slurmrestd - Improve handling when content plugins rely on parsers > that haven't been loaded. > -- Fix old pending jobs dying (Slurm version 21.08.x and older) when upgrading > Slurm due to "Invalid message version" errors. > -- Have client commands sleep for progressively longer periods when backed off > by the RPC rate limiting system. > -- slurmctld - Ensure agent queue is flushed correctly at shutdown time. > -- slurmdbd - correct lineage construction during assoc table conversion for > partition based associations. > -- Add new RPCs and API call for faster querying of job states from slurmctld. > -- slurmrestd - Add endpoint '/slurm/{data_parser}/jobs/state'. > -- squeue - Add `--only-job-state` argument to use faster query of job states. > -- Make a job requesting --no-requeue, or JobRequeue=0 in the slurm.conf, > supersede RequeueExit[Hold]. > -- Add sackd man page to the Debian package. > -- Fix issues with tasks when a job was shrinked more than once. > -- Fix reservation update validation that resulted in reject of correct > updates of reservation when the reservation was running jobs. > -- Fix possible segfault when the backup slurmctld is asserting control. > -- Fix regression introduced in 23.02.4 where slurmctld was not properly > tracking the total GRES selected for exclusive multi-node jobs, potentially > and incorrectly bypassing limits. > -- Fix tracking of jobs typeless GRES count when multiple typed GRES with the > same name are also present in the job allocation. Otherwise, the job could > bypass limits configured for the typeless GRES. > -- Fix tracking of jobs typeless GRES count when request specification has a > typeless GRES name first and then typed GRES of different names (i.e. > --gres=gpu:1,tmpfs:foo:2,tmpfs:bar:7). Otherwise, the job could bypass > limits configured for the generic of the typed one (tmpfs in the example). > -- Fix batch step not having SLURM_CLUSTER_NAME filled in. > -- slurmstepd - Avoid error during `--container` job cleanup about > RunTimeQuery never being configured. Results in cleanup where job steps not > fully started. > -- Fix nodes not being rebooted when using salloc/sbatch/srun "--reboot" flag. > -- Send scrun.lua in configless mode. > -- Fix rejecting an interactive job whose extra constraint request cannot > immediately be satisfied. > -- Fix regression in 23.11.0 when parsing LogTimeFormat=iso8601_ms that > prevented milliseconds from being printed. > -- Fix issue where you could have a gpu allocated as well as a shard on that > gpu allocated at the same time. > -- Fix slurmctld crashes when using extra constraints with job arrays. > -- sackd/slurmrestd/scrun - Avoid memory leak on new unix socket connection. > -- The failed node field is filled when a node fails but does not time out. > -- slurmrestd - Remove requiring job script field and job component script > fields to both be populated in the `POST /slurm/v0.0.40/job/submit` > endpoint as there can only be one batch step script for a job. > -- slurmrestd - When job script is provided in '.jobs[].script' and '.script' > fields, the '.script' field's value will be used in the > `POST /slurm/v0.0.40/job/submit` endpoint. > -- slurmrestd - Reject HetJob submission missing or empty batch script for > first Het component in the `POST /slurm/v0.0.40/job/submit` endpoint. > -- slurmrestd - Reject job when empty batch script submitted to the > POST /slurm/v0.0.40/job/submit` endpoint. > -- Fix pam_slurm and pam_slurm_adopt when using auth/slurm. > -- slurmrestd - Add 'cores_per_socket' field to > `POST /slurm/v0.0.40/job/submit` endpoint. > -- Fix srun and other Slurm commands running within a "configless" salloc when > salloc itself fetched the config. > -- Enforce binding with shared gres selection if requested. > -- Fix job allocation failures when the requested tres type or name ends in > "gres" or "license". > -- accounting_storage/mysql - Fix lineage string construction when adding a > user association with a partition. > -- Fix sattach command. > -- Fix ReconfigFlags. Due how reconfig was changed in 23.11, they will also > be used to influence the slurmctld startup as well. > -- Fix starting slurmd in configless mode if MUNGE support was disabled. -- Tim McMullan Release Management, Support, and Development SchedMD LLC - Commercial Slurm Development and Support

1 0

Re: Lua script
by Gestió Servidors 21 Mar '24

21 Mar '24

Hello, I answer about my question: * What is the contents of your /etc/slurm/job_submit.lua file? function slurm_job_submit(job_desc, part_list, submit_uid) if (job_desc.user_id == 1008) then slurm.log_info("Trabajo sometido por druiz") if (job_desc['partition'] == "nodo.q") then if (job_desc['time_limit'] > 345600) # 345600 seconds == 4 days # nodo.q partition has "PartitionName=nodo.q Nodes=clus[01-12] Default=YES MaxTime=04:00:00" configuration return slurm.FAILURE end end end return slurm.SUCCESS end slurm.log_info("initialized") return slurm.SUCCESS * Did you reconfigure slurmctld? Yes, in the first time, I ran "scontrol reconfigure", but after checking that limits wasn't applyed, I restarted slurmctld daemon * Check the log file by: grep job_submit /var/log/slurm/slurmctld.log In my slurmctld.log file in SLURM server, "grep job_submit /var/log/slurm/slurmctld.log" doesn't return anything... * What is your Slurm version? 23.11.0 Thanks.

1 0

Re: Lua script
by Gestió Servidors 20 Mar '24

20 Mar '24

Hello, after adding "EnforcePartLimits=ALL" in slurm.conf and restarting slurmctld daemon, job continues being accepted... so I don't undertand where I'm doing some wrong. My slurm.conf is this: ControlMachine=my_server MailProg=/bin/mail MpiDefault=none ProctrackType=proctrack/linuxproc ReturnToService=2 SlurmctldPidFile=/var/run/slurmctld.pid SlurmctldPort=6817 SlurmdPidFile=/var/run/slurmd.pid SlurmdPort=6818 SlurmdSpoolDir=/var/spool/slurmd SlurmUser=slurm SlurmdUser=root AuthType=auth/munge StateSaveLocation=/var/log/slurm SwitchType=switch/none TaskPlugin=task/none,task/affinity,task/cgroup TaskPluginParam=none DebugFlags=NO_CONF_HASH,Backfill,BackfillMap,SelectType,Steps,TraceJobs JobSubmitPlugins=lua SchedulerType=sched/backfill SelectType=select/cons_tres SelectTypeParameters=CR_Core SchedulerParameters=max_script_size=20971520 EnforcePartLimits=ALL CoreSpecPlugin=core_spec/none AccountingStorageType=accounting_storage/slurmdbd AccountingStoreFlags=job_comment JobCompType=jobcomp/filetxt JobCompLoc=/var/log/slurm/job_completions ClusterName=my_cluster JobAcctGatherType=jobacct_gather/linux SlurmctldDebug=5 SlurmctldLogFile=/var/log/slurmctld.log SlurmdDebug=5 SlurmdLogFile=/var/log/slurmd.log AccountingStorageEnforce=limits AccountingStorageHost=my_server NodeName=clus[01-06] CPUs=12 SocketsPerBoard=2 CoresPerSocket=6 ThreadsPerCore=1 RealMemory=128387 TmpDisk=81880 Feature=big-mem NodeName=clus[07-12] CPUs=12 SocketsPerBoard=2 CoresPerSocket=6 ThreadsPerCore=1 RealMemory=15491 TmpDisk=81880 Feature=small-mem NodeName=clus-login CPUs=4 SocketsPerBoard=2 CoresperSocket=2 ThreadsperCore=1 RealMemory=15886 TmpDisk=30705 PartitionName=nodo.q Nodes=clus[01-12] Default=YES MaxTime=04:00:00 State=UP AllocNodes=clus-login,clus05 MaxCPUsPerNode=12 KillOnBadExit=1 OverTimeLimit=30 # si el trabajo dura mas de 30 minutos despues del tiempo maximo (2 horas), se cancela TCPTimeout=5 PriorityType=priority/multifactor PriorityDecayHalfLife=7-0 PriorityCalcPeriod=5 PriorityUsageResetPeriod=QUARTERLY PriorityFavorSmall=NO PriorityMaxAge=7-0 PriorityWeightAge=10000 PriorityWeightFairshare=1000000 PriorityWeightJobSize=1000 PriorityWeightPartition=1000 PriorityWeightQOS=0 PropagateResourceLimitsExcept=MEMLOCK And testing script is this: #!/bin/bash #SBATCH --time=5-00:00:00 srun /bin/hostname date sleep 50 date Why my job is being submited into the queue and not refused BEFORE being queued? Thanks.

2 1

Node CPU Allocation and Effective CPU Allocation for Running Jobs
by Muhammad Akhdhor 19 Mar '24

19 Mar '24

Hi Everyone, We have a SLURM cluster of three different types of nodes. One partition consists of nodes that have a large number of CPUs, 256 CPUs on each node. I'm trying to find out the current CPU allocation on some of those nodes but part of the information I gathered seems to be incorrect. If I use "*scontrol show node <node-name>*", I get this for the CPU info: RealMemory=450000 AllocMem=262144 FreeMem=235397 Sockets=2 Boards=1 State=ALLOCATED ThreadsPerCore=2 TmpDisk=0 Weight=1 Owner=N/A MCS_label=N/A CPUAlloc=256 CPUEfctv=256 CPUTot=256 CPULoad=126.65 CfgTRES=cpu=256,mem=450000M,billing=256 AllocTRES=cpu=256,mem=256G However, when I tried to identify those jobs to which the node's CPUs have been allocated, and get a tally of the allocated CPUs, I can only see 128 CPUs that are effectively allocated on that node, based on the output of *squeue --state=R -o "%C %N".* So I don't quite understand why the running jobs on the nodes account for just 128, and not 256, CPU allocation even though scontrol reports 100% CPU allocation on the node. Could this be due to some misconfiguration, or a bug in the SLURM version we're running? We're running Version=23.02.4. The interesting thing is that we have six nodes that have similar specs, and all of them show up as allocated in the output of *sinfo*, but the running jobs on each node account for just 128 CPU allocation, as if they're all capped at 128. Any thoughts, suggestions or assistance to figure this out would be greatly appreciated. Thanks, Muhammad

2 2

Slurm suspend preemption not working
by Verma, Nischey (HPC ENG,RAL,LSCI) 16 Mar '24

16 Mar '24

Hi All, we are trying to implement preemption in one of our partitions so we can run priority jobs on it and suspend the ones running on the partition and resume once the priority job is done. We have read through the Slurm documentation and did the configuration, but somehow we can not make it work. We tried other preemption like cancel which works fine, but Suspend isn't working. The jobs with higher priority which should suspend other jobs stays as pending and waiting for resources. We are using a QOS to assign priority to the job and also configuring the QOS so it can preempt certain other QOSs. this is the output of our scontrol show config : AccountingStorageBackupHost = (null) AccountingStorageEnforce = associations,limits,qos AccountingStorageHost = -Some-server- AccountingStorageExternalHost = (null) AccountingStorageParameters = (null) AccountingStoragePort = 6819 AccountingStorageTRES = cpu,mem,energy,node,billing,fs/disk,vmem,pages AccountingStorageType = accounting_storage/slurmdbd AccountingStorageUser = N/A AccountingStoreFlags = job_comment,job_env,job_extra,job_script AcctGatherEnergyType = (null) AcctGatherFilesystemType = (null) AcctGatherInterconnectType = (null) AcctGatherNodeFreq = 0 sec AcctGatherProfileType = (null) AllowSpecResourcesUsage = No AuthAltTypes = (null) AuthAltParameters = (null) AuthInfo = (null) AuthType = auth/munge BatchStartTimeout = 10 sec BcastExclude = /lib,/usr/lib,/lib64,/usr/lib64 BcastParameters = (null) BOOT_TIME = 2024-03-15T13:32:05 BurstBufferType = (null) CliFilterPlugins = (null) ClusterName = cluster CommunicationParameters = (null) CompleteWait = 0 sec CoreSpecPlugin = (null) CpuFreqDef = Unknown CpuFreqGovernors = OnDemand,Performance,UserSpace CredType = cred/munge DebugFlags = (null) DefMemPerCPU = 3500 DependencyParameters = (null) DisableRootJobs = Yes EioTimeout = 60 EnforcePartLimits = ALL Epilog = (null) EpilogMsgTime = 2000 usec EpilogSlurmctld = (null) ExtSensorsType = (null) ExtSensorsFreq = 0 sec FairShareDampeningFactor = 1 FederationParameters = (null) FirstJobId = 1 GetEnvTimeout = 2 sec GresTypes = gpu GpuFreqDef = (null) GroupUpdateForce = 1 GroupUpdateTime = 600 sec HASH_VAL = Match HealthCheckInterval = 300 sec HealthCheckNodeState = ANY HealthCheckProgram = /usr/sbin/nhc InactiveLimit = 0 sec InteractiveStepOptions = --interactive --preserve-env --pty $SHELL JobAcctGatherFrequency = 30 JobAcctGatherType = jobacct_gather/linux JobAcctGatherParams = (null) JobCompHost = localhost JobCompLoc = (null) JobCompParams = (null) JobCompPort = 0 JobCompType = (null) JobCompUser = root JobContainerType = (null) JobDefaults = (null) JobFileAppend = 0 JobRequeue = 1 JobSubmitPlugins = lua KillOnBadExit = 0 KillWait = 30 sec LaunchParameters = (null) Licenses = (null) LogTimeFormat = iso8601_ms MailDomain = (null) MailProg = /bin/mail MaxArraySize = 1001 MaxBatchRequeue = 5 MaxDBDMsgs = 20024 MaxJobCount = 10000 MaxJobId = 67043328 MaxMemPerNode = UNLIMITED MaxNodeCount = 6 MaxStepCount = 40000 MaxTasksPerNode = 512 MCSPlugin = (null) MCSParameters = (null) MessageTimeout = 10 sec MinJobAge = 300 sec MpiDefault = pmix_v2 MpiParams = (null) NEXT_JOB_ID = 3626 NodeFeaturesPlugins = (null) OverTimeLimit = 0 min PluginDir = /usr/lib64/slurm PlugStackConfig = (null) PowerParameters = (null) PowerPlugin = (null) PreemptMode = GANG,SUSPEND PreemptParameters = (null) PreemptType = preempt/qos PreemptExemptTime = 00:00:00 PrEpParameters = (null) PrEpPlugins = prep/script PriorityParameters = (null) PrioritySiteFactorParameters = (null) PrioritySiteFactorPlugin = (null) PriorityDecayHalfLife = 7-00:00:00 PriorityCalcPeriod = 00:05:00 PriorityFavorSmall = No PriorityFlags = PriorityMaxAge = 7-00:00:00 PriorityUsageResetPeriod = NONE PriorityType = priority/multifactor PriorityWeightAge = 0 PriorityWeightAssoc = 0 PriorityWeightFairShare = 0 PriorityWeightJobSize = 0 PriorityWeightPartition = 0 PriorityWeightQOS = 5000 PriorityWeightTRES = (null) PrivateData = none ProctrackType = proctrack/cgroup Prolog = (null) PrologEpilogTimeout = 65534 PrologSlurmctld = (null) PrologFlags = (null) PropagatePrioProcess = 0 PropagateResourceLimits = ALL PropagateResourceLimitsExcept = (null) RebootProgram = /etc/slurm/slurmupdate.sh ReconfigFlags = (null) RequeueExit = (null) RequeueExitHold = (null) ResumeFailProgram = (null) ResumeProgram = (null) ResumeRate = 300 nodes/min ResumeTimeout = 60 sec ResvEpilog = (null) ResvOverRun = 0 min ResvProlog = (null) ReturnToService = 1 SchedulerParameters = bf_max_job_user=2 SchedulerTimeSlice = 30 sec SchedulerType = sched/backfill ScronParameters = (null) SelectType = select/cons_tres SelectTypeParameters = CR_CORE_MEMORY SlurmUser = slurm(888) SlurmctldAddr = (null) SlurmctldDebug = info SlurmctldHost[0] = -some-server- SlurmctldLogFile = /var/log/slurm/slurmctld.log SlurmctldPort = 6817 SlurmctldSyslogDebug = (null) SlurmctldPrimaryOffProg = (null) SlurmctldPrimaryOnProg = (null) SlurmctldTimeout = 120 sec SlurmctldParameters = (null) SlurmdDebug = info SlurmdLogFile = /var/log/slurm/slurmd.log SlurmdParameters = (null) SlurmdPidFile = /var/run/slurmd.pid SlurmdPort = 6818 SlurmdSpoolDir = /var/spool/slurm SlurmdSyslogDebug = (null) SlurmdTimeout = 300 sec SlurmdUser = root(0) SlurmSchedLogFile = (null) SlurmSchedLogLevel = 0 SlurmctldPidFile = /var/run/slurmctld.pid SLURM_CONF = /etc/slurm/slurm.conf SLURM_VERSION = 23.11.1 SrunEpilog = (null) SrunPortRange = 0-0 SrunProlog = (null) StateSaveLocation = /var/spool/slurmctld SuspendExcNodes = (null) SuspendExcParts = (null) SuspendExcStates = (null) SuspendProgram = (null) SuspendRate = 60 nodes/min SuspendTime = INFINITE SuspendTimeout = 30 sec SwitchParameters = (null) SwitchType = (null) TaskEpilog = (null) TaskPlugin = task/cgroup,task/affinity TaskPluginParam = none TaskProlog = (null) TCPTimeout = 2 sec TmpFS = /tmp TopologyParam = (null) TopologyPlugin = topology/default TrackWCKey = No TreeWidth = 16 UsePam = No UnkillableStepProgram = (null) UnkillableStepTimeout = 60 sec VSizeFactor = 0 percent WaitTime = 0 sec X11Parameters = (null) Cgroup Support Configuration: AllowedRAMSpace = 100.0% AllowedSwapSpace = 0.0% CgroupMountpoint = /sys/fs/cgroup CgroupPlugin = autodetect ConstrainCores = yes ConstrainDevices = yes ConstrainRAMSpace = yes ConstrainSwapSpace = no EnableControllers = no IgnoreSystemd = no IgnoreSystemdOnFailure = no MaxRAMPercent = 100.0% MaxSwapPercent = 100.0% MemorySwappiness = (null) MinRAMSpace = 30 MB MPI Plugins Configuration: PMIxCliTmpDirBase = (null) PMIxCollFence = (null) PMIxDebug = 0 PMIxDirectConn = yes PMIxDirectConnEarly = no PMIxDirectConnUCX = no PMIxDirectSameArch = no PMIxEnv = (null) PMIxFenceBarrier = no PMIxNetDevicesUCX = (null) PMIxTimeout = 300 PMIxTlsUCX = (null) the partition is configured like: PartitionName=test-partition AllowGroups=sysadmin,users AllowAccounts=ALL AllowQos=ALL AllocNodes=ALL Default=NO QoS=N/A DefaultTime=NONE DisableRootJobs=YES ExclusiveUser=NO GraceTime=0 Hidden=NO MaxNodes=UNLIMITED MaxTime=UNLIMITED MinNodes=0 LLN=NO MaxCPUsPerNode=UNLIMITED MaxCPUsPerSocket=UNLIMITED Nodes=vserv-[275-277] PriorityJobFactor=1 PriorityTier=100 RootOnly=NO ReqResv=NO OverSubscribe=FORCE:1 OverTimeLimit=NONE PreemptMode=GANG,SUSPEND State=UP TotalCPUs=32 TotalNodes=4 SelectTypeParameters=NONE JobDefaults=(null) DefMemPerNode=UNLIMITED MaxMemPerNode=UNLIMITED TRES=cpu=32,mem=304933M,node=4,billing=32 Our QOS looks like this: | Name | Priority | GraceTime | Preempt | PreemptMode | Flags | UsageFactor | MaxTRESPU | MaxJobsPU | MaxSubmitPU |-----------|----------|-----------|-----------|-------------|-----------------|-------------|-----------------|-----------|----------- | normal | 50 | 00:00:00 | | cluster | | 1.000000 | | 20 | 50 | preempter | 1000 | 00:00:00 | preempted | gang,suspe+| | 1.000000 | | | | preempted | 0 | 00:00:00 | | gang,suspe+ | OverPartQOS | 1.000000 | cpu=100,node=10 | | I can provide more configs if needed. Do you guys see anything strange? Or any property to be set? This is the state of the queue: |JOBID | QOS | ST | TIME | NODELIST(REASON) | PARTITION | PRIORITY | | ---------------- ----------- --------- ----------- ---------------------- ----------------- ----------| | 3629 | preempted | PD | 0:00 | (Resources) | test-partition | 1 | | 3627 | preempted | R | 0:20 | vserv-276 | test-partition | 1 | | 3628 | preempted | R | 0:20 | vserv-277 | test-partition | 1 | | 3626 | preempted | R | 0:27 | vserv-275 | test-partition | 1 | | 3630 | preempter | PD | 0:00 | (Resources) | test-partition | 5000 | any advice is welcomed. Regards, Nischey Verma Nischey Verma

3 2

Issues with Slurm 23.11.1
by Fokke Dijkstra 16 Mar '24

16 Mar '24

Dear all, Since the upgrade from Slurm 22.05 to 23.11.1 we are having problems with the communication between the slurmctld and slurmd processes. We are running a cluster with 183 nodes and almost 19000 cores. Unfortunately some nodes are in a different network preventing full internode communication. A network topology and setting TopologyParam RouteTree have been used to make sure no slurmd communication happens between nodes on different networks. In the new Slurm version we see the following issues, which did not appear in 22.05: 1. slurmd processes acquire many network connections in CLOSE-WAIT (or CLOSE_WAIT depending on the tool used) causing the processes to hang, when trying to restart slurmd. When checking for CLOSE-WAIT processes we see the following behaviour: Recv-Q Send-Q Local Address:Port Peer Address:Port Process 1 0 10.5.2.40:6818 10.5.0.43:58572 users:(("slurmd",pid=1930095,fd=72),("slurmd",pid=1930067,fd=72)) 1 0 10.5.2.40:6818 10.5.0.43:58284 users:(("slurmd",pid=1930095,fd=8),("slurmd",pid=1930067,fd=8)) 1 0 10.5.2.40:6818 10.5.0.43:58186 users:(("slurmd",pid=1930095,fd=22),("slurmd",pid=1930067,fd=22)) 1 0 10.5.2.40:6818 10.5.0.43:58592 users:(("slurmd",pid=1930095,fd=76),("slurmd",pid=1930067,fd=76)) 1 0 10.5.2.40:6818 10.5.0.43:58338 users:(("slurmd",pid=1930095,fd=19),("slurmd",pid=1930067,fd=19)) 1 0 10.5.2.40:6818 10.5.0.43:58568 users:(("slurmd",pid=1930095,fd=68),("slurmd",pid=1930067,fd=68)) 1 0 10.5.2.40:6818 10.5.0.43:58472 users:(("slurmd",pid=1930095,fd=69),("slurmd",pid=1930067,fd=69)) 1 0 10.5.2.40:6818 10.5.0.43:58486 users:(("slurmd",pid=1930095,fd=38),("slurmd",pid=1930067,fd=38)) 1 0 10.5.2.40:6818 10.5.0.43:58316 users:(("slurmd",pid=1930095,fd=29),("slurmd",pid=1930067,fd=29)) The first IP address is that of the compute node, the second that of the node running slurmctld. The nodes can communicate using these IP addresses just fine. 2. slurmd cannot be properly restarted [2024-01-18T10:45:26.589] slurmd version 23.11.1 started [2024-01-18T10:45:26.593] error: Error binding slurm stream socket: Address already in use [2024-01-18T10:45:26.593] fatal: Unable to bind listen port (6818): Address already in use This is probably because of the processes being in CLOSE-WAIT, which can only be killed using signal -9. 3. We see jobs stuck in completing CG state, probably due to communication issues between slurmctld and slurmd. The slurmctld sends repeated kill requests but those do not seem to be acknowledged by the client. This happens more often in large job arrays, or generally when many jobs start at the same time. However, this could be just a biased observation (i.e., it is more noticeable on large job arrays because there are more jobs to fail in the first place). 4. Since the new version we also see messages like: [2024-01-17T09:58:48.589] error: Failed to kill program loading user environment [2024-01-17T09:58:48.590] error: Failed to load current user environment variables [2024-01-17T09:58:48.590] error: _get_user_env: Unable to get user's local environment, running only with passed environment The effect of this is that the users run with the wrong environment and can’t load the modules for the software that is needed by their jobs. This leads to many job failures. The issue appears to be somewhat similar to the one described at: https://bugs.schedmd.com/show_bug.cgi?id=18561 In that case the site downgraded the slurmd clients to 22.05 which got rid of the problems. We’ve now downgraded the slurmd on the compute nodes to 23.02.7 which also seems to be a workaround for the issue. Does anyone know of a better solution? Kind regards, Fokke Dijkstra -- Fokke Dijkstra <f.dijkstra(a)rug.nl> <f.dijkstra(a)rug.nl> Team High Performance Computing Center for Information Technology, University of Groningen Postbus 11044, 9700 CA Groningen, The Netherlands

2 3

Overhead of multiple concurrent job steps
by Mehta, Kshitij 15 Mar '24

15 Mar '24

Hello, We have a use case in which we need to launch multiple concurrently running MPI applications inside a job allocation. Most supercomputing facilities limit the number of concurrent job steps as they incur an overhead with the global Slurm scheduler. Some frameworks, such as the Flux framework from LLNL, claim to mitigate this issue by starting an instance of their own scheduler inside an allocation, which then acts as the resource manager for the compute nodes in the allocation. Out of curiosity, I was wondering if there is a fundamental reason behind having a single global scheduler that the srun launch commands must contact to launch job steps. Perhaps it was overkill to develop a ‘hierarhical’ design in which Slurm launches a local job daemon for every allocation that manages resources for that allocation? I would appreciate your insight in understanding more about Slurm’s core design. Thanks and regards, Kshitij Mehta Oak Ridge National Laboratory

1 0

Jobs being denied for GrpCpuLimit despite having enough resource
by Simon Andrews 15 Mar '24

15 Mar '24

Our cluster has developed a strange intermittent behaviour where jobs are being put into a pending state because they aren't passing the AssocGrpCpuLimit, even though the user submitting has enough cpus for the job to run. For example: $ squeue -o "%.6i %.9P %.8j %.8u %.2t %.10M %.7m %.7c %.20R" JOBID PARTITION NAME USER ST TIME MIN_MEM MIN_CPU NODELIST(REASON) 799 normal hostname andrewss PD 0:00 2G 5 (AssocGrpCpuLimit) ..so the job isn't running, and it's the only job in the queue, but: $ sacctmgr list associations part=normal user=andrewss format=Account,User,Partition,Share,GrpTRES Account User Partition Share GrpTRES ---------- ---------- ---------- --------- ------------- andrewss andrewss normal 1 cpu=5 That user has a limit of 5 CPUs so the job should run. The weird thing is that this effect is intermittent. A job can hang and the queue will stall for ages but will then suddenly start working and you can submit several jobs and they all work, until one fails again. The cluster has active nodes and plenty of resource: $ sinfo PARTITION AVAIL TIMELIMIT NODES STATE NODELIST normal* up infinite 2 idle compute-0-[6-7] interactive up 1-12:00:00 3 idle compute-1-[0-1,3] The slurmctld log just says: [2024-03-14T16:21:41.275] _slurm_rpc_submit_batch_job: JobId=799 InitPrio=4294901720 usec=259 Whilst it's in this state I can run other jobs with core requests of up to 4 and they work, but not 5. It's like slurm is adding one CPU to the request and then denying it. I'm sure I'm missing something fundamental but would appreciate it if someone could point out what it is! Thanks Simon.

3 3

2025

2024