--- /dev/null
+openstack:
+ - volumes: # attached to each instance
+ count: 3
+ size: 30 # GB
--- /dev/null
+meta:
+- desc: |
+ Run ceph on two nodes,
+ with a separate client 0,1,2 third node.
+ Use xfs beneath the osds.
+ CephFS tests running on client 2,3
+roles:
+- - mon.a
+ - mgr.x
+ - mds.a
+ - osd.0
+ - osd.1
+- - mon.b
+ - mon.c
+ - osd.2
+ - osd.3
+- - client.0
+ - client.1
+ - client.2
+ - client.3
+overrides:
+ ceph:
+ log-whitelist:
+ - scrub mismatch
+ - ScrubResult
+ - wrongly marked
+ - (POOL_APP_NOT_ENABLED)
+ - overall HEALTH_
+ conf:
+ global:
+ enable experimental unrecoverable data corrupting features: "*"
+ mon:
+ mon warn on osd down out interval zero: false
+ osd:
+ osd_class_load_list: "cephfs hello journal lock log numops rbd refcount
+ replica_log rgw sdk statelog timeindex user version"
+ osd_class_default_list: "cephfs hello journal lock log numops rbd refcount
+ replica_log rgw sdk statelog timeindex user version"
+ fs: xfs
--- /dev/null
+meta:
+- desc: |
+ install ceph/luminous latest
+ run workload and upgrade-sequence in parallel
+ upgrade the client node
+tasks:
+- install:
+ branch: luminous
+- print: "**** done installing luminous"
+- ceph:
+ log-whitelist:
+ - overall HEALTH_
+ - \(FS_
+ - \(MDS_
+ - \(OSD_
+ - \(MON_DOWN\)
+ - \(CACHE_POOL_
+ - \(POOL_
+ - \(MGR_DOWN\)
+ - \(PG_
+ - \(SMALLER_PGP_NUM\)
+ - Monitor daemon marked osd
+ - Behind on trimming
+ - Manager daemon
+ conf:
+ global:
+ mon warn on pool no app: false
+- exec:
+ osd.0:
+ - ceph osd require-osd-release luminous
+ - ceph osd set-require-min-compat-client luminous
+- print: "**** done ceph"
+- install.upgrade:
+ mon.a:
+ mon.b:
+- print: "**** done install.upgrade both hosts"
+- parallel:
+ - workload
+ - upgrade-sequence
+- print: "**** done parallel"
+- install.upgrade:
+ client.0:
+- print: "**** done install.upgrade on client.0"
--- /dev/null
+meta:
+- desc: |
+ run a cephfs stress test
+ mount ceph-fuse on client.2 before running workunit
+workload:
+ full_sequential:
+ - sequential:
+ - ceph-fuse:
+ - print: "**** done ceph-fuse 2-workload"
+ - workunit:
+ clients:
+ client.2:
+ - suites/blogbench.sh
+ - print: "**** done suites/blogbench.sh 2-workload"
--- /dev/null
+meta:
+- desc: |
+ run run randomized correctness test for rados operations
+ on an erasure-coded pool
+workload:
+ full_sequential:
+ - rados:
+ clients: [client.0]
+ ops: 4000
+ objects: 50
+ ec_pool: true
+ write_append_excl: false
+ op_weights:
+ read: 100
+ write: 0
+ append: 100
+ delete: 50
+ snap_create: 50
+ snap_remove: 50
+ rollback: 50
+ copy_from: 50
+ setattr: 25
+ rmattr: 25
+ - print: "**** done rados ec task"
--- /dev/null
+meta:
+- desc: |
+ object class functional tests
+workload:
+ full_sequential:
+ - workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - cls
+ - print: "**** done cls 2-workload"
--- /dev/null
+meta:
+- desc: |
+ generate read/write load with rados objects ranging from 1MB to 25MB
+workload:
+ full_sequential:
+ - workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - rados/load-gen-big.sh
+ - print: "**** done rados/load-gen-big.sh 2-workload"
--- /dev/null
+meta:
+- desc: |
+ librbd C and C++ api tests
+workload:
+ full_sequential:
+ - workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - rbd/test_librbd.sh
+ - print: "**** done rbd/test_librbd.sh 2-workload"
--- /dev/null
+meta:
+- desc: |
+ librbd python api tests
+workload:
+ full_sequential:
+ - workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - rbd/test_librbd_python.sh
+ - print: "**** done rbd/test_librbd_python.sh 2-workload"
--- /dev/null
+meta:
+- desc: |
+ upgrade the ceph cluster
+upgrade-sequence:
+ sequential:
+ - ceph.restart:
+ daemons: [mon.a, mon.b, mon.c, mgr.x]
+ - ceph.restart:
+ daemons: [osd.0, osd.1, osd.2, osd.3]
+ wait-for-healthy: false
+ wait-for-osds-up: true
+ - ceph.restart:
+ daemons: [mds.a]
+ wait-for-healthy: false
+ wait-for-osds-up: true
+ - print: "**** done ceph.restart all"
--- /dev/null
+meta:
+- desc: |
+ upgrade the ceph cluster,
+ upgrate in two steps
+ step one ordering: mon.a, osd.0, osd.1, mds.a
+ step two ordering: mon.b, mon.c, osd.2, osd.3
+ ceph expected to be healthy state after each step
+upgrade-sequence:
+ sequential:
+ - ceph.restart:
+ daemons: [mon.a]
+ wait-for-healthy: true
+ - sleep:
+ duration: 60
+ - ceph.restart:
+ daemons: [mon.b, mon.c, mgr.x]
+ wait-for-healthy: true
+ - sleep:
+ duration: 60
+ - ceph.restart:
+ daemons: [osd.0, osd.1]
+ wait-for-healthy: true
+ - sleep:
+ duration: 60
+ - ceph.restart: [mds.a]
+ - sleep:
+ duration: 60
+ - sleep:
+ duration: 60
+ - ceph.restart:
+ daemons: [osd.2, osd.3]
+ wait-for-healthy: false
+ wait-for-osds-up: true
+ - sleep:
+ duration: 60
--- /dev/null
+meta:
+- desc: |
+ run a cephfs stress test
+ mount ceph-fuse on client.3 before running workunit
+tasks:
+- sequential:
+ - ceph-fuse:
+ - print: "**** done ceph-fuse 5-final-workload"
+ - workunit:
+ clients:
+ client.3:
+ - suites/blogbench.sh
+ - print: "**** done suites/blogbench.sh 5-final-workload"
--- /dev/null
+meta:
+- desc: |
+ randomized correctness test for rados operations on a replicated pool with snapshots
+tasks:
+ - rados:
+ clients: [client.1]
+ ops: 4000
+ objects: 50
+ write_append_excl: false
+ op_weights:
+ read: 100
+ write: 100
+ delete: 50
+ snap_create: 50
+ snap_remove: 50
+ rollback: 50
+ - print: "**** done rados 4-final-workload"
--- /dev/null
+meta:
+- desc: |
+ generate read/write load with rados objects ranging from 1 byte to 1MB
+tasks:
+ - workunit:
+ clients:
+ client.1:
+ - rados/load-gen-mix.sh
+ - print: "**** done rados/load-gen-mix.sh 4-final-workload"
--- /dev/null
+meta:
+- desc: |
+ librados C and C++ api tests
+overrides:
+ ceph:
+ log-whitelist:
+ - reached quota
+tasks:
+ - mon_thrash:
+ revive_delay: 20
+ thrash_delay: 1
+ - print: "**** done mon_thrash 4-final-workload"
+ - workunit:
+ branch: luminous
+ clients:
+ client.1:
+ - rados/test.sh
+ - print: "**** done rados/test.sh 4-final-workload"
--- /dev/null
+meta:
+- desc: |
+ rbd object class functional tests
+tasks:
+ - workunit:
+ clients:
+ client.1:
+ - cls/test_cls_rbd.sh
+ - print: "**** done cls/test_cls_rbd.sh 4-final-workload"
--- /dev/null
+meta:
+- desc: |
+ run basic import/export cli tests for rbd
+tasks:
+ - workunit:
+ clients:
+ client.1:
+ - rbd/import_export.sh
+ env:
+ RBD_CREATE_ARGS: --new-format
+ - print: "**** done rbd/import_export.sh 4-final-workload"
--- /dev/null
+meta:
+- desc: |
+ swift api tests for rgw
+overrides:
+ rgw:
+ frontend: civetweb
+tasks:
+ - rgw: [client.1]
+ - print: "**** done rgw 4-final-workload"
+ - swift:
+ client.1:
+ rgw_server: client.1
+ - print: "**** done swift 4-final-workload"
--- /dev/null
+../../../../distros/supported/
\ No newline at end of file
--- /dev/null
+../stress-split/objectstore/
\ No newline at end of file
--- /dev/null
+../stress-split/0-cluster/
\ No newline at end of file
--- /dev/null
+../stress-split/2-partial-upgrade/
\ No newline at end of file
--- /dev/null
+meta:
+- desc: |
+ randomly kill and revive osd
+ small chance to increase the number of pgs
+overrides:
+ ceph:
+ log-whitelist:
+ - but it is still running
+ - wrongly marked me down
+ - objects unfound and apparently lost
+ - log bound mismatch
+tasks:
+- parallel:
+ - stress-tasks
+stress-tasks:
+- thrashosds:
+ timeout: 1200
+ chance_pgnum_grow: 1
+ chance_pgpnum_fix: 1
+ min_in: 4
+ chance_thrash_cluster_full: 0
+ chance_thrash_pg_upmap: 0
+ chance_thrash_pg_upmap_items: 0
+ chance_force_recovery: 0
+- print: "**** done thrashosds 3-thrash"
--- /dev/null
+meta:
+- desc: |
+ randomized correctness test for rados operations on an erasure coded pool
+stress-tasks:
+ - rados:
+ clients: [client.0]
+ ops: 4000
+ objects: 50
+ ec_pool: true
+ write_append_excl: false
+ op_weights:
+ read: 100
+ write: 0
+ append: 100
+ delete: 50
+ snap_create: 50
+ snap_remove: 50
+ rollback: 50
+ copy_from: 50
+ setattr: 25
+ rmattr: 25
+ - print: "**** done rados ec task"
--- /dev/null
+../stress-split/5-finish-upgrade.yaml
\ No newline at end of file
--- /dev/null
+#
+# k=3 implies a stripe_width of 1376*3 = 4128 which is different from
+# the default value of 4096 It is also not a multiple of 1024*1024 and
+# creates situations where rounding rules during recovery becomes
+# necessary.
+#
+meta:
+- desc: |
+ randomized correctness test for rados operations on an erasure coded pool
+ using the jerasure plugin with k=3 and m=1
+tasks:
+- rados:
+ clients: [client.0]
+ ops: 4000
+ objects: 50
+ ec_pool: true
+ write_append_excl: false
+ erasure_code_profile:
+ name: jerasure31profile
+ plugin: jerasure
+ k: 3
+ m: 1
+ technique: reed_sol_van
+ crush-failure-domain: osd
+ op_weights:
+ read: 100
+ write: 0
+ append: 100
+ delete: 50
+ snap_create: 50
+ snap_remove: 50
+ rollback: 50
+ copy_from: 50
+ setattr: 25
+ rmattr: 25
--- /dev/null
+../../../../distros/supported/
\ No newline at end of file
--- /dev/null
+../stress-split/objectstore/
\ No newline at end of file
--- /dev/null
+../../../../tasks/thrashosds-health.yaml
\ No newline at end of file
--- /dev/null
+openstack:
+ - machine:
+ disk: 100 # GB
+ - volumes: # attached to each instance
+ count: 3
+ size: 30 # GB
--- /dev/null
+meta:
+- desc: |
+ Run ceph on two nodes,
+ with a separate client-only node.
+ Use xfs beneath the osds.
+overrides:
+ ceph:
+ fs: xfs
+ log-whitelist:
+ - overall HEALTH_
+ - \(MON_DOWN\)
+ - \(MGR_DOWN\)
+ conf:
+ global:
+ enable experimental unrecoverable data corrupting features: "*"
+ mon:
+ mon warn on osd down out interval zero: false
+roles:
+- - mon.a
+ - mon.b
+ - mon.c
+ - mgr.x
+ - osd.0
+ - osd.1
+ - osd.2
+- - osd.3
+ - osd.4
+ - osd.5
+- - client.0
--- /dev/null
+meta:
+- desc: install ceph/luminous latest
+tasks:
+- install:
+ branch: luminous
+- print: "**** done install luminous"
+- ceph:
+- exec:
+ osd.0:
+ - ceph osd require-osd-release luminous
+ - ceph osd set-require-min-compat-client luminous
+- print: "**** done ceph "
+overrides:
+ ceph:
+ conf:
+ mon:
+ mon warn on osd down out interval zero: false
--- /dev/null
+meta:
+- desc: |
+ install upgrade ceph/-x on one node only
+ 1st half
+ restart : osd.0,1,2
+tasks:
+- install.upgrade:
+ osd.0:
+- print: "**** done install.upgrade osd.0"
+- ceph.restart:
+ daemons: [mon.a,mon.b,mon.c,mgr.x,osd.0,osd.1,osd.2]
+- print: "**** done ceph.restart 1st half"
--- /dev/null
+meta:
+- desc: |
+ randomly kill and revive osd
+ small chance to increase the number of pgs
+overrides:
+ ceph:
+ log-whitelist:
+ - but it is still running
+ - wrongly marked me down
+ - objects unfound and apparently lost
+ - log bound mismatch
+tasks:
+- parallel:
+ - stress-tasks
+stress-tasks:
+- thrashosds:
+ timeout: 1200
+ chance_pgnum_grow: 1
+ chance_pgpnum_fix: 1
+ chance_thrash_cluster_full: 0
+ chance_thrash_pg_upmap: 0
+ chance_thrash_pg_upmap_items: 0
+ disable_objectstore_tool_tests: true
+ chance_force_recovery: 0
+- print: "**** done thrashosds 3-thrash"
--- /dev/null
+meta:
+- desc: |
+ run randomized correctness test for rados operations
+ generate write load with rados bench
+stress-tasks:
+- full_sequential:
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+ - radosbench:
+ clients: [client.0]
+ time: 150
+- print: "**** done radosbench 7-workload"
--- /dev/null
+meta:
+- desc: |
+ run basic cls tests for rbd
+stress-tasks:
+- workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - cls/test_cls_rbd.sh
+- print: "**** done cls/test_cls_rbd.sh 5-workload"
--- /dev/null
+meta:
+- desc: |
+ run basic import/export cli tests for rbd
+stress-tasks:
+- workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - rbd/import_export.sh
+ env:
+ RBD_CREATE_ARGS: --new-format
+- print: "**** done rbd/import_export.sh 5-workload"
--- /dev/null
+meta:
+- desc: |
+ librbd C and C++ api tests
+stress-tasks:
+- workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - rbd/test_librbd.sh
+- print: "**** done rbd/test_librbd.sh 7-workload"
--- /dev/null
+meta:
+- desc: |
+ randomized correctness test for rados operations on a replicated pool,
+ using only reads, writes, and deletes
+stress-tasks:
+- full_sequential:
+ - rados:
+ clients: [client.0]
+ ops: 4000
+ objects: 500
+ write_append_excl: false
+ op_weights:
+ read: 45
+ write: 45
+ delete: 10
+- print: "**** done rados/readwrite 5-workload"
--- /dev/null
+meta:
+- desc: |
+ randomized correctness test for rados operations on a replicated pool with snapshot operations
+stress-tasks:
+- full_sequential:
+ - rados:
+ clients: [client.0]
+ ops: 4000
+ objects: 50
+ write_append_excl: false
+ op_weights:
+ read: 100
+ write: 100
+ delete: 50
+ snap_create: 50
+ snap_remove: 50
+ rollback: 50
+- print: "**** done rados/snaps-few-objects 5-workload"
--- /dev/null
+tasks:
+- install.upgrade:
+ osd.3:
+ client.0:
+- ceph.restart:
+ daemons: [osd.3, osd.4, osd.5]
+ wait-for-healthy: false
+ wait-for-osds-up: true
+
--- /dev/null
+meta:
+- desc: |
+ librbd python api tests
+tasks:
+- workunit:
+ branch: luminous
+ clients:
+ client.0:
+ - rbd/test_librbd_python.sh
+- print: "**** done rbd/test_librbd_python.sh 9-workload"
--- /dev/null
+meta:
+- desc: |
+ swift api tests for rgw
+tasks:
+- rgw:
+ client.0:
+- print: "**** done rgw 9-workload"
+- swift:
+ client.0:
+ rgw_server: client.0
+- print: "**** done swift 9-workload"
--- /dev/null
+meta:
+- desc: |
+ randomized correctness test for rados operations on a replicated pool with snapshot operations
+tasks:
+- rados:
+ clients: [client.0]
+ ops: 4000
+ objects: 500
+ write_append_excl: false
+ op_weights:
+ read: 100
+ write: 100
+ delete: 50
+ snap_create: 50
+ snap_remove: 50
+ rollback: 50
--- /dev/null
+../../../../distros/supported/
\ No newline at end of file
--- /dev/null
+../../../../../objectstore/bluestore.yaml
\ No newline at end of file
--- /dev/null
+../../../../../objectstore/filestore-xfs.yaml
\ No newline at end of file
--- /dev/null
+../../../../tasks/thrashosds-health.yaml
\ No newline at end of file