Skip to content

crush: add multistep retry rules#55096

Merged
yuriw merged 19 commits intoceph:mainfrom
athanatos:sjust/for-review/wip-crush-msr
Jan 26, 2024
Merged

crush: add multistep retry rules#55096
yuriw merged 19 commits intoceph:mainfrom
athanatos:sjust/for-review/wip-crush-msr

Conversation

@athanatos
Copy link
Contributor

@athanatos athanatos commented Jan 8, 2024

Checklist

  • Tracker (select at least one)
    • References tracker ticket
    • Very recent bug; references commit where it was introduced
    • [x ] New feature (ticket optional)
    • [ x] Doc update (no ticket needed)
    • Code cleanup (no ticket needed)
  • Component impact
    • Affects Dashboard, opened tracker ticket
    • Affects Orchestrator, opened tracker ticket
    • [ x] No impact that needs to be tracked
  • Documentation (select at least one)
    • [x ] Updates relevant documentation
    • No doc update is appropriate
  • Tests (select at least one)
Show available Jenkins commands
  • jenkins retest this please
  • jenkins test classic perf
  • jenkins test crimson perf
  • jenkins test signed
  • jenkins test make check
  • jenkins test make check arm64
  • jenkins test submodules
  • jenkins test dashboard
  • jenkins test dashboard cephadm
  • jenkins test api
  • jenkins test docs
  • jenkins render docs
  • jenkins test ceph-volume all
  • jenkins test ceph-volume tox
  • jenkins test windows
  • jenkins test rook e2e

@athanatos
OSDMonitor: update require_osd_release for crush features
ec9fdd4 
The existing implementation checks that the currently live
OSDs support the pending crush features, but we don't
actually check that require_osd_release is new enough.

Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
src/crush: add a rule type enum
22ab274 
We're going to add more rule types that don't map directly onto
pool types.

Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
crush: add msr rule types and CrushWrapper infrastructure
196205d 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
osd/OSDMap: minor formatting nit
5265455 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos athanatos requested review from a team as code owners January 8, 2024 19:23
@athanatos
Copy link
Contributor Author

athanatos commented Jan 8, 2024

Added thrash-erasure-code big tests pass https://pulpito.ceph.com/sjust-2024-01-04_18:17:17-rados:thrash-erasure-code-big-wip-sjust-msr-testing-2024-01-03-distro-default-smithi/

@athanatos
Copy link
Contributor Author

athanatos commented Jan 8, 2024

Checking/fixing make check failures -- related to tool output differences

athanatos added 14 commits January 8, 2024 19:25
@athanatos
crush: add msr tunables
88a7fc2 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
crush: add choosemsr rule
d822572 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
crush/mapper: add support for MSR types
2c7b6bb 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
mon/OSDMonitor: generalize rule type check for pools
1199166 
Add rule_valid_for_pool_type to CrushWrapper to generalize
rule type <-> pool type mapping to include the new MSR
types.

Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
test/crush: convert indep test cases to test MSR as well
9adedac 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
test/crush: add test variants for firstn rules
6276009 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
test/crush: add tests specifically for MSR
845d91d 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
vstart.sh: add --osds-per-host
791a7b1 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
vstart.sh: add --require-osd-and-client-version and --use-crush-tunab…
1441634 
…les flags

Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
erasure-code: add support for multiple osds in a single failure domain
404e84f 
Adds support for crush-osds-per-failure-domain and
crush-num-failure-domains via MSR rules.

Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
doc/rados/operations: add CRUSH MSR documentation
e6a3b55 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
qa/erasure-code: modify jerasure 4/2 ec test case to use msr
67e71ae 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
test/cli/crushtool/choose-args.t: add msr related json output
72996a9 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos
test/cli/osdmaptool/crush.t: adjust --import-crush size output
38bd19a 
Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos athanatos force-pushed the sjust/for-review/wip-crush-msr branch from 0c08d59 to 38bd19a Compare January 9, 2024 03:27
@athanatos
tasks/.../test_erasure_code_profile: assertSubset in test_create_plugin
ef46191 
Newly added profile options may break this test otherwise.

Signed-off-by: Samuel Just <sjust@redhat.com>
@athanatos athanatos requested a review from a team as a code owner January 9, 2024 19:36
@athanatos athanatos requested review from afreen23 and ivoalmeida and removed request for a team January 9, 2024 19:36
@athanatos athanatos requested a review from ljflores January 12, 2024 20:48
@ljflores ljflores requested review from ljflores and removed request for ljflores January 12, 2024 20:48
ljflores
ljflores reviewed Jan 12, 2024
View reviewed changes
doc/rados/operations/crush-map-edits.rst
step emit
}

CRUSH MSR rules are a distinct type of CRUSH rule which supports retrying steps
Copy link
Member

@ljflores ljflores Jan 12, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
CRUSH MSR rules are a distinct type of CRUSH rule which supports retrying steps
CRUSH MSR (multi-step retry) rules are a distinct type of CRUSH rule which supports retrying steps

@athanatos athanatos requested a review from jdurgin January 17, 2024 22:16
athanatos
athanatos commented Jan 17, 2024
View reviewed changes
src/crush/mapper.c
int **step_vecs;
};

/// encapsulates output space, invariant through crush_msr_do_rule invocation
Copy link
Contributor Author

@athanatos athanatos Jan 17, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

invariant isn't the right word -- the output space remains the same but is mutated as items are output (out[i] and returned_so_far). TODO

@athanatos
Copy link
Contributor Author

athanatos commented Jan 22, 2024

jenkins test make check arm64

@anthonyeleven
Copy link
Contributor

anthonyeleven commented Jan 22, 2024

Suggest a few sentences in the PR description.

@ljflores
Copy link
Member

ljflores commented Jan 26, 2024

Rados-approved: https://tracker.ceph.com/projects/rados/wiki/MAIN#httpstrellocomctbflL3Ja1936-wip-yuri3-testing-2024-01-22-1155

@athanatos
Copy link
Contributor Author

athanatos commented Jan 26, 2024

Let's hold of on merging for now, I'll take care of merging it.

@yuriw yuriw merged commit 37d5d93 into ceph:main Jan 26, 2024
@athanatos
Copy link
Contributor Author

athanatos commented Jan 26, 2024

Once the revert merges, I'll repost this PR.

@athanatos athanatos mentioned this pull request Jan 26, 2024
Merged
14 tasks
@athanatos
Copy link
Contributor Author

athanatos commented Jan 26, 2024

#55332 is the new iteration of this PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ljflores ljflores ljflores approved these changes

@gregsfortytwo gregsfortytwo Awaiting requested review from gregsfortytwo

@benhanokh benhanokh Awaiting requested review from benhanokh

@ivoalmeida ivoalmeida Awaiting requested review from ivoalmeida ivoalmeida is a code owner automatically assigned from ceph/dashboard

@afreen23 afreen23 Awaiting requested review from afreen23 afreen23 is a code owner automatically assigned from ceph/dashboard

@jdurgin jdurgin Awaiting requested review from jdurgin

Assignees

No one assigned

Labels

core dashboard documentation mon needs-qa tests wip-yuri3-testing

Projects

Ceph-Dashboard
Archived in project

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@athanatos @anthonyeleven @ljflores @yuriw