From 89d695fb8b684086a1c13b5727ddba6e201d439d Mon Sep 17 00:00:00 2001 From: Nitzan Mordechai Date: Wed, 26 Jun 2024 08:01:09 +0000 Subject: [PATCH] suites: check for host thrasher The last PR modified the suites to only check for host thrasher. This update fixes that issue by implementing different settings with dedicated YAML files for host thrashing Fixes: https://tracker.ceph.com/issues/66657 Signed-off-by: Nitzan Mordechai --- qa/suites/rados/thrash/thrashers/careful.yaml | 1 - .../rados/thrash/thrashers/careful_host.yaml | 28 +++++++++++++++++ qa/suites/rados/thrash/thrashers/default.yaml | 1 - .../rados/thrash/thrashers/default_host.yaml | 30 ++++++++++++++++++ .../rados/thrash/thrashers/mapgap_host.yaml | 31 +++++++++++++++++++ qa/suites/rados/thrash/thrashers/pggrow.yaml | 1 - .../rados/thrash/thrashers/pggrow_host.yaml | 26 ++++++++++++++++ 7 files changed, 115 insertions(+), 3 deletions(-) create mode 100644 qa/suites/rados/thrash/thrashers/careful_host.yaml create mode 100644 qa/suites/rados/thrash/thrashers/default_host.yaml create mode 100644 qa/suites/rados/thrash/thrashers/mapgap_host.yaml create mode 100644 qa/suites/rados/thrash/thrashers/pggrow_host.yaml diff --git a/qa/suites/rados/thrash/thrashers/careful.yaml b/qa/suites/rados/thrash/thrashers/careful.yaml index 7a866fc009d0..47d2db48169f 100644 --- a/qa/suites/rados/thrash/thrashers/careful.yaml +++ b/qa/suites/rados/thrash/thrashers/careful.yaml @@ -21,7 +21,6 @@ tasks: - thrashosds: timeout: 1200 min_in: 2 - thrash_hosts: true chance_pgnum_grow: 1 chance_pgnum_shrink: 1 chance_pgpnum_fix: 1 diff --git a/qa/suites/rados/thrash/thrashers/careful_host.yaml b/qa/suites/rados/thrash/thrashers/careful_host.yaml new file mode 100644 index 000000000000..7a866fc009d0 --- /dev/null +++ b/qa/suites/rados/thrash/thrashers/careful_host.yaml @@ -0,0 +1,28 @@ +overrides: + ceph: + log-ignorelist: + - but it is still running + - objects unfound and apparently lost + conf: + osd: + osd debug reject backfill probability: .3 + osd scrub min interval: 60 + osd scrub max interval: 120 + osd max backfills: 3 + osd snap trim sleep: 2 + mon: + mon min osdmap epochs: 50 + paxos service trim min: 10 + # prune full osdmaps regularly + mon osdmap full prune min: 15 + mon osdmap full prune interval: 2 + mon osdmap full prune txsize: 2 +tasks: +- thrashosds: + timeout: 1200 + min_in: 2 + thrash_hosts: true + chance_pgnum_grow: 1 + chance_pgnum_shrink: 1 + chance_pgpnum_fix: 1 + aggressive_pg_num_changes: false diff --git a/qa/suites/rados/thrash/thrashers/default.yaml b/qa/suites/rados/thrash/thrashers/default.yaml index 6a0cf59a2afd..592c77800463 100644 --- a/qa/suites/rados/thrash/thrashers/default.yaml +++ b/qa/suites/rados/thrash/thrashers/default.yaml @@ -22,7 +22,6 @@ tasks: - thrashosds: timeout: 1200 min_in: 2 - thrash_hosts: true chance_pgnum_grow: 1 chance_pgnum_shrink: 1 chance_pgpnum_fix: 1 diff --git a/qa/suites/rados/thrash/thrashers/default_host.yaml b/qa/suites/rados/thrash/thrashers/default_host.yaml new file mode 100644 index 000000000000..6a0cf59a2afd --- /dev/null +++ b/qa/suites/rados/thrash/thrashers/default_host.yaml @@ -0,0 +1,30 @@ +overrides: + ceph: + log-ignorelist: + - but it is still running + - objects unfound and apparently lost + conf: + osd: + osd debug reject backfill probability: .3 + osd scrub min interval: 60 + osd scrub max interval: 120 + osd max backfills: 3 + osd snap trim sleep: 2 + osd delete sleep: 1 + mon: + mon min osdmap epochs: 50 + paxos service trim min: 10 + # prune full osdmaps regularly + mon osdmap full prune min: 15 + mon osdmap full prune interval: 2 + mon osdmap full prune txsize: 2 +tasks: +- thrashosds: + timeout: 1200 + min_in: 2 + thrash_hosts: true + chance_pgnum_grow: 1 + chance_pgnum_shrink: 1 + chance_pgpnum_fix: 1 + chance_bluestore_reshard: 1 + bluestore_new_sharding: random diff --git a/qa/suites/rados/thrash/thrashers/mapgap_host.yaml b/qa/suites/rados/thrash/thrashers/mapgap_host.yaml new file mode 100644 index 000000000000..707eaa97492a --- /dev/null +++ b/qa/suites/rados/thrash/thrashers/mapgap_host.yaml @@ -0,0 +1,31 @@ +overrides: + ceph: + log-ignorelist: + - but it is still running + - objects unfound and apparently lost + - osd_map_cache_size + conf: + mon: + mon min osdmap epochs: 50 + paxos service trim min: 10 + # prune full osdmaps regularly + mon osdmap full prune min: 15 + mon osdmap full prune interval: 2 + mon osdmap full prune txsize: 2 + osd: + osd map cache size: 1 + osd scrub min interval: 60 + osd scrub max interval: 120 + osd scrub during recovery: false + osd max backfills: 6 + osd beacon report interval: 30 +tasks: +- thrashosds: + timeout: 1800 + min_in: 2 + thrash_hosts: true + chance_pgnum_grow: 0.25 + chance_pgnum_shrink: 0.25 + chance_pgpnum_fix: 0.25 + chance_test_map_discontinuity: 2 + map_discontinuity_sleep_time: 200 diff --git a/qa/suites/rados/thrash/thrashers/pggrow.yaml b/qa/suites/rados/thrash/thrashers/pggrow.yaml index d3c5a63d0d22..79f96fd0a8d2 100644 --- a/qa/suites/rados/thrash/thrashers/pggrow.yaml +++ b/qa/suites/rados/thrash/thrashers/pggrow.yaml @@ -21,6 +21,5 @@ tasks: - thrashosds: timeout: 1200 min_in: 2 - thrash_hosts: true chance_pgnum_grow: 2 chance_pgpnum_fix: 1 diff --git a/qa/suites/rados/thrash/thrashers/pggrow_host.yaml b/qa/suites/rados/thrash/thrashers/pggrow_host.yaml new file mode 100644 index 000000000000..d3c5a63d0d22 --- /dev/null +++ b/qa/suites/rados/thrash/thrashers/pggrow_host.yaml @@ -0,0 +1,26 @@ +overrides: + ceph: + log-ignorelist: + - but it is still running + - objects unfound and apparently lost + conf: + osd: + osd scrub min interval: 60 + osd scrub max interval: 120 + filestore odsync write: true + osd max backfills: 2 + osd snap trim sleep: .5 + mon: + mon min osdmap epochs: 50 + paxos service trim min: 10 + # prune full osdmaps regularly + mon osdmap full prune min: 15 + mon osdmap full prune interval: 2 + mon osdmap full prune txsize: 2 +tasks: +- thrashosds: + timeout: 1200 + min_in: 2 + thrash_hosts: true + chance_pgnum_grow: 2 + chance_pgpnum_fix: 1 -- 2.47.3