slurm_job_terminate core dump

[laparn]

Hello to all,

I am trying on slurm 2.5 slurm_job_terminate. What I am doing :

import pyslurm
jobs=pyslurm.job()
dod=jobs.get()
dod
{174: {u'comment': None, u'time_limit': 26L, u'cnode_cnt': None, u'alloc_node': u'D3550', u'features': [], u'eligible_time': 1357308185, u'contiguous': False, u'resv_id': None, u'ramdisk_image': None, u'block_id': None, u'sockets_per_node': 65534, u'req_switch': 0L, u'resv_name': None, u'licenses': {}, u'qos': None, u'submit_time': 1357308185, u'mloader_image': None, u'num_cpus': 1L, u'conn_type': (None, 'None'), u'show_flags': 0, u'user_id': 1001L, u'network': None, u'restart_cnt': 0, u'work_dir': u'/home/arnaud/src/slurmjob', u'pn_min_tmp_disk': 0L, u'max_nodes': 0L, u'job_state': (1, 'RUNNING'), u'assoc_id': 0L, u'exit_code': 0L, u'num_nodes': 1L, u'priority': 4294901721L, u'batch_script': None, u'boards_per_node': 0, u'ntasks_per_socket': 65535, u'batch_flag': 0, u'derived_ec': 0L, u'nodes': None, u'preempt_time': 0, u'pn_min_cpus': 1, u'nice': 10000, u'ntasks_per_node': 0, u'linux_image': None, u'altered': None, u'sockets_per_board': 0, u'alloc_sid': 10014L, u'start_time': 1357308185, u'pre_sus_time': 0, u'ionodes': None, u'state_reason': (0, 'None'), u'pn_min_memory': 0L, u'rotate': False, u'reboot': None, u'blrts_image': None, u'shared': 0, u'time_min': 10L, u'wait4switch': 0L, u'ntasks_per_core': 65535, u'wckey': None, u'account': None, u'requeue': True, u'name': u'wait-arg.sh', u'req_nodes': [], u'gres': [], u'suspend_time': 0, u'partition': 'debug', u'cores_per_socket': 65534, u'batch_host': u'D3550', u'dependency': None, u'max_cpus': 0L, u'state_desc': None, u'command': u'/home/arnaud/src/slurmjob/./wait-arg.sh', u'end_time': 1357309745, u'cpus_per_task': 1, u'resize_time': 0, u'group_id': 1001L, u'exc_nodes': [], u'threads_per_core': 65534}}
pyslurm.pyslurm.slurm_terminate_job(174)
Erreur de segmentation (core dumped)

Strange as at the same time notify, kill, suspend and resume are perfectly working.

[gingergeeks]

Another one for us to investigate, what platform/cluster are you doing this on ?

Thanks for testing things out !

Mark

[gingergeeks]

I've been able to reproduce and I'm going through the slurm source to trace the issue

[gingergeeks]

I wrote a very small C program to call the same function for a jobstep (slurm-2.5.0) which generated a core. Using gdb I get

Program terminated with signal 11, Segmentation fault.
#0  0x00007fef6075631a in select_g_select_jobinfo_pack (jobinfo=, buffer=0x618578,
    protocol_version=6400) at node_select.c:1010
1010                    pack32(*(ops[plugin_id].plugin_id), buffer);

So it looks and smells of a Slurm issue, I will test slurm-2.5.1 as well and raise a ticket or chat with Danny and Moe.

[laparn]

Nice. Thanks for having traced it.