Merge pull request ceph#38891 from yuriw/wip-yuriw-octopus-x-master

qa/tests: added stress-split-erasure-code-no-cephadm tests

Reviewed-by: Neha Ojha <[email protected]>
Reviewed-by: Josh Durgin <[email protected]>

qa/suites/upgrade/octopus-x/parallel/workload/rbd_import_export.yaml

@@ -1,14 +1,14 @@
 meta:
-- desc: |
-   run basic import/export cli tests for rbd
-workload:
-  full_sequential:
-    - print: "**** done start rbd_import_export.yaml"
-    - workunit:
-        branch: octopus
-        clients:
-          client.1:
-              - rbd/import_export.sh
-        env:
-          RBD_CREATE_ARGS: --new-format
-    - print: "**** done end rbd_import_export.yaml"
+- desc: |
+   run basic import/export cli tests for rbd
+workload:
+  full_sequential:
+    - print: "**** done start rbd_import_export.yaml"
+    - workunit:
+        branch: octopus
+        clients:
+          client.1:
+              - rbd/import_export.sh
+        env:
+          RBD_CREATE_ARGS: --new-format
+    - print: "**** done end rbd_import_export.yaml"

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/.qa

@@ -0,0 +1 @@
+../.qa

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/0-cluster

@@ -0,0 +1 @@
+../stress-split-no-cephadm/0-cluster

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/1-nautilus-install

@@ -0,0 +1 @@
+../stress-split-no-cephadm/1-ceph-install

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/1.1-pg-log-overrides/.qa

@@ -0,0 +1 @@
+../.qa/

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/1.1-pg-log-overrides/normal_pg_log.yaml

@@ -0,0 +1 @@
+

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/1.1-pg-log-overrides/short_pg_log.yaml

@@ -0,0 +1,6 @@
+overrides:
+  ceph:
+    conf:
+      osd:
+        osd min pg log entries: 1
+        osd max pg log entries: 2

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/2-partial-upgrade

@@ -0,0 +1 @@
+../stress-split-no-cephadm/2-partial-upgrade

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/3-thrash/.qa

@@ -0,0 +1 @@
+../.qa/

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/3-thrash/default.yaml

@@ -0,0 +1,26 @@
+meta:
+- desc: |
+   randomly kill and revive osd
+   small chance to increase the number of pgs
+overrides:
+  ceph:
+    log-ignorelist:
+    - 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
+    aggressive_pg_num_changes: false
+- print: "**** done thrashosds default.yaml"

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/3.1-objectstore

@@ -0,0 +1 @@
+../stress-split-no-cephadm/objectstore

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/4-ec-workload/.qa

@@ -0,0 +1 @@
+../.qa/

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/4-ec-workload/rados-ec-workload.yaml

@@ -0,0 +1,22 @@
+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"

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/4-ec-workload/rbd-ec-workload.yaml

@@ -0,0 +1,31 @@
+meta:
+- desc: |
+   run rbd tests on EC pool
+   overrides => force bluestore since it's required for ec-overwrite
+   use an EC pool for rbd and run xfstests on top of it to verify correctness
+tasks:
+- exec:
+    client.0:
+      - sudo ceph osd erasure-code-profile set teuthologyprofile crush-failure-domain=osd m=1 k=2
+      - sudo ceph osd pool create datapool 4 4 erasure teuthologyprofile
+      - sudo ceph osd pool set datapool allow_ec_overwrites true
+      - rbd pool init datapool
+- qemu:
+    all:
+      clone: true
+      type: block
+      disks: 3
+      test: qa/run_xfstests_qemu.sh
+- print: "**** done rbd/qemu ec task"
+exclude_arch: armv7l
+overrides:
+  thrashosds:
+    bdev_inject_crash: 2
+    bdev_inject_crash_probability: .5
+  ceph:
+    fs: xfs
+    conf:
+      client:
+        rbd default data pool: datapool
+      osd: # force bluestore since it's required for ec overwrites
+        osd objectstore: bluestore

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/5-finish-upgrade.yaml

@@ -0,0 +1 @@
+../stress-split-no-cephadm/5-finish-upgrade.yaml

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/6-pacific.yaml

@@ -0,0 +1 @@
+.qa/releases/pacific.yaml

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/7-final-workload.yaml

@@ -0,0 +1,36 @@
+#
+# 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
+- print: "**** done rados ec 7-final-workload.yaml"

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/mon_election

@@ -0,0 +1 @@
+.qa/mon_election

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/thrashosds-health.yaml

@@ -0,0 +1 @@
+.qa/tasks/thrashosds-health.yaml

qa/suites/upgrade/octopus-x/stress-split-erasure-code-no-cephadm/ubuntu_18.04.yaml

@@ -0,0 +1 @@
+.qa/distros/all/ubuntu_18.04.yaml

qa/suites/upgrade/octopus-x/stress-split-no-cephadm/4-workload/radosbench.yaml

@@ -49,4 +49,4 @@ stress-tasks:
   - radosbench:
       clients: [client.0]
       time: 90
-- print: "**** done radosbench 7-workload"
+- print: "**** done end radosbench.yaml"

qa/suites/upgrade/octopus-x/stress-split-no-cephadm/5-finish-upgrade.yaml

@@ -11,11 +11,11 @@ tasks:
     daemons: [mon.c, osd.8, osd.9, osd.10, osd.11, rgw.*]
     wait-for-healthy: false
     wait-for-osds-up: true
-- print: "**** restarted/upgrated => mon.c, osd.8, osd.9, osd.10, osd.11, rgw.*"
+- print: "**** done restarted/upgrated => mon.c, osd.8, osd.9, osd.10, osd.11, rgw.*"
 - exec:
     osd.0:
       - ceph osd set pglog_hardlimit
       - ceph osd dump --format=json-pretty | grep "flags"
       - ceph config set global mon_warn_on_msgr2_not_enabled false
-- print: "**** try to set pglog_hardlimit again, should succeed"
+- print: "**** done try to set pglog_hardlimit again, should succeed"
 

qa/suites/upgrade/octopus-x/stress-split-no-cephadm/8-final-workload/rbd-python.yaml

@@ -11,4 +11,4 @@ tasks:
     clients:
       client.0:
         - rbd/test_librbd_python.sh
-- print: "**** done rbd/test_librbd_python.sh 9-workload"
+- print: "**** done rbd/test_librbd_python.sh"