The distinct hardware/software components that a lab is composed of and the way
they interact between them is described here. In general, a lab is composed of
-a scheduler (a.k.a. master node), worker(s), package builder
-(`gitbuilder <http://ceph.com/gitbuilder.cgi>`__), job database (`paddles
+a scheduler, worker(s), package builder (`gitbuilder
+<http://ceph.com/gitbuilder.cgi>`__), job database (`paddles
<https://github.com/ceph/paddles>`__), job archive, a web UI (`pulpito
<https://github.com/ceph/pulpito>`__) and test nodes.
UI. For an example, see Ceph community's Lab `here <http://pulpito.ceph.com>`__.
Test nodes can be grouped in classes (referred to as ``machine-type``),
-allowing teuthology schedule jobs across multiple hardware setups,
-provided they're visible to the master node.
+allowing teuthology schedule jobs across multiple hardware setups.
Life of a Teuthology Job
========================
is created (or whenever the status of a job is changed, e.g. from queued to
started), information about the job is recorded in ``paddles``'s internal
database. Depending on the priority of the job, the scheduler eventually
-determines when a job can get executed. At this point, the master communicates
-with ``teuthology-dispatcher``, checks the lock status of the requested
+determines when a job can get executed. At this point,
+``teuthology-dispatcher`` checks the lock status of the requested
machines by querying ``paddles``, acquires locks of the
nodes if they are available, and invokes ``teuthology-dispatcher`` in
``supervisor`` mode. ``supervisor`` reimages the target machines and invokes
}
},
'flavor': Placeholder('flavor'),
- 'log-whitelist': ['\(MDS_ALL_DOWN\)',
- '\(MDS_UP_LESS_THAN_MAX\)'],
'log-ignorelist': ['\(MDS_ALL_DOWN\)',
'\(MDS_UP_LESS_THAN_MAX\)'],
'sha1': Placeholder('ceph_hash'),
'debug': {'mds_client.yaml': None},
'fs': {'btrfs.yaml': None},
'msgr-failures': {'none.yaml': None},
- 'overrides': {'whitelist_wrongly_marked_down.yaml': None},
+ 'overrides': {'allowlist_wrongly_marked_down.yaml': None},
'tasks': {'cfuse_workunit_suites_fsstress.yaml': None},
},
}
fail for other denials one can add the overrides with appropriate escapes
overrides:
selinux:
- whitelist:
+ allowlist:
- 'name="cephtest"'
- 'dmidecode'
- 'comm="logrotate"'
'comm="sss_cache"',
'context=system_u:system_r:NetworkManager_dispatcher_t:s0',
]
- se_whitelist = self.config.get('whitelist', [])
- if se_whitelist:
- known_denials.extend(se_whitelist)
+ se_allowlist = self.config.get('allowlist', [])
+ if se_allowlist:
+ known_denials.extend(se_allowlist)
ignore_known_denials = '\'\(' + str.join('\|', known_denials) + '\)\''
for remote in self.cluster.remotes.keys():
proc = remote.run(