cephadm: share code between 'pull' and 'inspect-image'

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: upgrade: pull image after upgrade start, and for each host

Make 'upgrade start' return quickly, without first pulling the image.

Pull the image once to establish the image_id.

For each host, before updating the container, ensure the local image is
up to date, and if not, pull.

If a pull returns a different image_id, restart upgrade process.  (This
could live-lock if two hosts have different container registries that
return different image ids for the same image name. :/)

Signed-off-by: Sage Weil <sage@redhat.com>

cephadm: add inspect-image command

This has the same output as pull, except that it inspects the copy of the
image that is local, and does not do a pull.

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: get --osd-uuid deploy arg from osdmap

During the initial OSD creation, we had the osd_uuid handy.  However, on
reconfig/redeploy, we don't.

Refactor a bit so that this is extracted from the osdmap by the
_create_daemon() method.  While we're at it, make the deployment case
verify that the deployed OSD uuid matches the one in the osdmap.

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: fetch keyring in _create_daemon()

This simplifies life for the callers, especially service_action
reconfig and redeploy.

Signed-off-by: Sage Weil <sage@redhat.com>

Merge pull request #32797 from tchaikov/wip-seastar-and-errorator

seastar,crimson: pickup change to pin socket to fixed core

Reviewed-by: Ronen Friedman <rfriedma@redhat.com>
Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

Merge pull request #29703 from baixueyu/wip-quota

rgw:  when you abort a multipart upload request, the quota may be not updated

Merge pull request #28622 from IlsooByun/master

rgw: incorrect return value when processing CORS headers

Merge pull request #29733 from baixueyu/wip-reload-coredump

rgw: maybe coredump when reload operator happened

Merge pull request #29795 from xxcs/xxcs03

rgw: fix one part of the bulk delete(RGWDeleteMultiObj_ObjStore_S3) fails but no error messages

Merge pull request #30913 from ofriedma/wip-ofriedma-lc-logs

rgw: Add more details to the LC delete and transit log

Merge PR #32787 into master

* refs/pull/32787/head:
qa/tasks/cephadm: work around .git suffix on ceph_repo
qa/tasks/cephadm: learn to pull cephadm from github

Reviewed-by: Jan Fajerski <jfajerski@suse.com>

mgr/dashboard: Add column filtering feature to cd-table (#32568)

mgr/dashboard: Add column filtering feature to cd-table

Reviewed-by: Stephan Müller <smueller@suse.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>

Merge pull request #32387 from athanatos/sjust/wip-glibcxx-assertions

Enable _GLIBCXX_ASSERTIONS and fix unittest problems

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Casey Bodley <cbodley@redhat.com>

qa/tasks/cephadm: work around .git suffix on ceph_repo

githubusercontent.com doesn't like the .git suffix

Signed-off-by: Sage Weil <sage@redhat.com>

Merge PR #32794 into master

* refs/pull/32794/head:
cephadm: normalize image_id in 'pull' output too
cephadm: fix image_id normalization
cephadm: pull: remove newline from ceph_version
mgr/cephadm: base upgrade check on return code, not stderr
mgr/cephadm: fix ok-to-stop
mgr/cephadm: some upgrade debug output
mgr/cephadm: remove dup check
mgr/cephadm: only need_upgrade_self if image_id doesn't match
mgr/cephadm: safe to restart mgr and other non-mon/osd/mds daemons

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge pull request #32791 from liewegas/fix-rgw-website

qa/suites/rgw/website: run test on ubuntu

Merge pull request #32276 from smithfarm/wip-drop-repo-access-doc

doc/cleanup: drop repo-access.rst

Reviewed-by: Zac Dover <zac.dover@gmail.com>

Merge pull request #32269 from liuchang0812/issue-43199-lockdep-in-coroutines-manager

rgw: get elastic search info in start_sync, avoid creating new coroutines manager

Merge pull request #32290 from jiahuizeng/es-valid

rgw: relax es zone validity check

Merge pull request #32458 from ofriedma/wip-ofriedma-fix-trans-noncurr

cls/rgw: when object is versioned and lc transition it, the object is becoming non-current

Merge pull request #32180 from yehudasa/wip-43255

rgw: use bucket creation time from bucket instance info

Merge pull request #32736 from capri1989/credits-script

src/script/credits.sh - switch to bash

Reviewed-By: Abhishek Lekshmanan <abhishek@suse.com>

crimson/common/errorator: construct future<> with a future_state_base

this change is made to adapt
https://github.com/scylladb/seastar/commit/89e801b373423b068b3a70e50b656348f549ade0

* always use the helpers from `futurize` instead of using
  `::seastar::make_exception_future()` directly. as
  `make_exception_future(future_state_base&& state)` is located in
  `seastar::internal` namespace
* add `noexcept` specifier in both constructors. to match with the
  helpers of `make_exception_future2()` which dispatch the call to
  different constructor variants.

Signed-off-by: Kefu Chai <kchai@redhat.com>

seastar: pickup change to pin socket to fixed core

and clang / GCC-10 related fixes

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #32373 from zdover23/wip-doc-pg-formats-update-dec-2019

doc: update list of formats for --format flag for "ceph pg dump"

Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #32790 from neha-ojha/wip-pglog-defaults

doc/rados: update osd_min_pg_log_entries and add osd_max_pg_log_entries

Reviewed-by: Sage Weil <sage@redhat.com>

rgw_kms: fix concat_url for empty url case

Can't call back() on an empty string.  Causes unittest_rgw_kms
to fail when compiled with _GLIBCXX_ASSERTIONS.

Signed-off-by: Samuel Just <sjust@redhat.com>

fastbmap_allocator_impl: avoid std::vector[] for oob index

This was relatively harmless as the pointer won't actually be
dereferenced, but it runs afoul of std::vector's bounds checks with
_GLIBCXX_ASSERTIONS enabled.

Signed-off-by: Samuel Just <sjust@redhat.com>

unittest_random: min > max triggers a glibc assert

Signed-off-by: Samuel Just <sjust@redhat.com>

src/CMakeLists.txt: enable _GLIBCXX_ASSERTIONS in debug builds

We already enable this by default in our rpm builds.  Adding it here as
well increases the odds we'll catch things with make check and local
testing.

Signed-off-by: Samuel Just <sjust@redhat.com>

cephadm: normalize image_id in 'pull' output too

Signed-off-by: Sage Weil <sage@redhat.com>

cephadm: fix image_id normalization

Remove sha256: prefix from id (hash), not name.

Signed-off-by: Sage Weil <sage@redhat.com>

Merge PR #32741 into master

* refs/pull/32741/head:
qa/cephfs: make filelock_interrupt.py work with python3

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

cephadm: pull: remove newline from ceph_version

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: base upgrade check on return code, not stderr

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: fix ok-to-stop

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: some upgrade debug output

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: remove dup check

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: only need_upgrade_self if image_id doesn't match

Signed-off-by: Sage Weil <sage@redhat.com>

mgr/cephadm: safe to restart mgr and other non-mon/osd/mds daemons

We don't have "ok-to-stop" checks for these.  And the mgr check for
self/active is already in the caller.

Signed-off-by: Sage Weil <sage@redhat.com>

qa/suites/rgw/website: run test on ubuntu

The dnsmasq package on centos 8.0 is broken, see
https://tracker.ceph.com/issues/43744

For now, run this test on ubuntu.

Signed-off-by: Sage Weil <sage@redhat.com>

Merge pull request #30152 from theanalyst/rgw/iam-policy-bitset-fn

rgw: iam: use a function to calculate the Action Bit string

Reviewed-by: Adam C. Emerson <aemerson@redhat.com>

Merge pull request #29851 from linuxbox2/wip-rgw-linkvtrack

rgw: fix version tracking across bucket link steps

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #30540 from mikulely/uri-beast-cleanup

rgw: fix REQUEST_URI setting in the rgw_asio_client.cc

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #30539 from mikulely/fix-operation2

rgw: fix opslog operation field as per Amazon s3

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #30471 from liuchang0812/delete-bucket-with-system-user

rgw: the http response code of delete bucket should not be 204-no-content

Reviewed-by: Casey Bodley <cbodley@redhat.com>

doc/rados/configuration: document osd_max_pg_log_entries

Signed-off-by: Neha Ojha <nojha@redhat.com>

doc/rados: update osd_min_pg_log_entries

update docs to match 0db140c15c16ce45de2e59f075de2b70c0a9cd95

Signed-off-by: Neha Ojha <nojha@redhat.com>

Merge PR #32644 into master

* refs/pull/32644/head:
qa: ignore trimmed cache items for dead cache drop
qa: use unit test comparisons

Reviewed-by: Zheng Yan <zyan@redhat.com>

Merge pull request #32772 from tchaikov/wip-43403

test/unittest_lockdep: do not start extra threads

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #31356 from xxcs/xxcs06

rgw: Incorrectly calling ceph::buffer::list::decode_base64 in bucket policy

Reviewed-by: Adam C. Emerson <aemerson@redhat.com>

Merge pull request #31355 from zhangsw/fix-rgw-reshard-max-entries

rgw: reshard list may return more than specified max_entries.

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #31344 from zhangsw/fix-rgw-reshard-lock-failed

rgw: return error if lock log shard fails

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #31239 from joke-lee/rgw-modify-zone-placement

rgw: radosgw-admin zone placement rm and radosgw-admin zonegroup placement rm support --storage-class

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #31591 from liuchang0812/current-version-in-rgw-user-info

rgw: fix RGWUserInfo decode current version

Reviewed-by: Casey Bodley <cbodley@redhat.com>
Reviewed-by: Daniel Gryniewicz <dang@redhat.com>

qa/tasks/cephadm: learn to pull cephadm from github

Fixes: https://tracker.ceph.com/issues/43755
Signed-off-by: Sage Weil <sage@redhat.com>

Merge PR #30636 into master

* refs/pull/30636/head:
common: Fix multiple logical errors in get_device_id.

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge PR #32732 into master

* refs/pull/32732/head:
msg/async: use dedicated method for clearing conn's priv up.
msg/async: mark down local_connection before draining the stack.

Reviewed-by: Kefu Chai <kchai@redhat.com>

googletest: pick up change to print GTEST_SKIP messages

to include the fix for
https://github.com/google/googletest/pull/2517

Signed-off-by: Kefu Chai <kchai@redhat.com>

test/unittest_lockdep: do not start extra threads

also, stop linking against libglobal

Fixes: https://tracker.ceph.com/issues/43403
Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge PR #32726 into master

* refs/pull/32726/head:
mon/ConfigMonitor: clean out bad config records
mon/ConfigMonitor: make 'config get' on fsid work
common/options: remove NO_MON_UPDATE from mon_dns_srv_name
mon/ConfigMonitor: do not 'config get' on NO_MON_UPDATE options
mon/ConfigMonitor: do not set NO_MON_UPDATE values

Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>

Merge pull request #32699 from tspmelo/wip-npm-audit-fix

mgr/dashboard: Fix npm vulnerabilities

Reviewed-by: Alfonso Martínez <almartin@redhat.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>

Merge pull request #32653 from votdev/issue_43603_clone_role

mgr/dashboard: Clone an existing user role

Reviewed-by: Tiago Melo <tmelo@suse.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>

Merge pull request #32648 from p-se/wip-pse-cd-api-err-msg

mgr/dashboard: monitoring: improve generic "Could not reach external API" message

Reviewed-by: Tatjana Dehler <tdehler@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>

Merge PR #32765 into master

* refs/pull/32765/head:
unittest_lockdep: avoid any threads for death test

Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #32728 from rishabh-d-dave/doc-teuth-broken-links

doc: fix links in developer_guide

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #32703 from s0nea/wip-dashboard-pwd-expiration-field

mgr/dashboard: fix visibility of pwdExpirationDate field

Reviewed-by: Volker Theile <vtheile@suse.com>
Reviewed-by: Stephan Müller <smueller@suse.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>

Merge pull request #32683 from liewegas/wip-variable-pg-log

osd: trim pg logs based on a per-osd budget

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

Merge pull request #32770 from tchaikov/wip-gtest-skip

test/librados: use GTEST_SKIP() to skip test

Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge PR #31719 into master

* refs/pull/31719/head:
mds: Reorganize class members in Server header

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #31232 into master

* refs/pull/31232/head:
test: test case for openfiletable MAX_ITEMS_PER_OBJ value verification
mds/OpenFileTable: match MAX_ITEMS_PER_OBJ to osd_deep_scrub_large_omap_object_key_threshold

Reviewed-by: Zheng Yan <zyan@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #31817 into master

* refs/pull/31817/head:
mds: send delegated_inos in openc response
mds: make take_ino just return 0 if it can't find one
mds: add infrastructure for delegating inos from the prealloc set
mds: fix dump of prealloc and used inos
mds: don't warn when inodes are not allocated out sequentially
mds: fix decoding version for EMetaBlob
mds: remove MDRequestImpl::did_ino_allocation()
mds: add a comment explaining what session_info_t is
doc: fix awkward phrasing in caching.txt

Reviewed-by: Zheng Yan <zyan@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #32326 into master

* refs/pull/32326/head:
mds: Reorganize class members in SnapClient header

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #32706 into master

* refs/pull/32706/head:
mount.ceph: remove arbitrary limit on size of name= option

Reviewed-by: Ilya Dryomov <idryomov@redhat.com>

test/librados: use GTEST_SKIP() to skip test

the output would look like

```
[  SKIPPED ] SkipTest.DoesSkip (0 ms)
```

easier for tester/developer to review the test result.

Signed-off-by: Kefu Chai <kchai@redhat.com>

osd/PeeringState: maximize log entries sent during backfill

Send as much log as we can during backfill so that dup op detection works
as well as it can.

Signed-off-by: Sage Weil <sage@redhat.com>

osd/PeeringState: always use target pg log length

Previously, the min and max log length was the same, so we didn't adjust
our log length based on whether we were degraded.  Preserve that behavior
by removing the degraded check entirely.  This keeps memory usage
consistent when recovery happens.

Signed-off-by: Sage Weil <sage@redhat.com>

osd: trim pg logs based on a per-osd budget

Set the default budget based on the current defaults: 3000 per osd, and a
rule of thumb target of 100 PGs per OSD.  Set the per-PG trim target
by dividing the overall value by the number of PGs on the OSD.

Increase the max pg log length alone, so if the OSD has <100 PGs,
those PGs will get more entries.  Reduce the minimum to be smaller than
the max.  Use the min/max config options to bracket what is allocated to
a single PG.

Signed-off-by: Sage Weil <sage@redhat.com>

crimson/osd: drop unused member variable

Signed-off-by: Kefu Chai <kchai@redhat.com>

crimson/osd: track number of pgs in shard_services

pg will be using it for evaluating the number pg log entries to keep for
each PG

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge PR #32522 into master

* refs/pull/32522/head:
doc/cephfs: improve wording in mount-prerequisites.rst
doc: migrate best practices recommendations to relevant docs
doc/cephfs: rename doc/cephfs/kernel.rst & doc/cephfs/fuse.rst

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #32763 into master

* refs/pull/32763/head:
qa: build v5.4 kernel

Reviewed-by: Jeff Layton <jlayton@redhat.com>

Merge PR #32762 into master

* refs/pull/32762/head:
doc/cephfs/client-auth: description and example are inconsistent

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #32739 into master

* refs/pull/32739/head:
qa: log warning on scrub error

Reviewed-by: Neha Ojha <nojha@redhat.com>

unittest_lockdep: avoid any threads for death test

Heed the warning:

[WARNING] /home/sage/src/ceph/src/googletest/googletest/src/gtest-death-test.cc:1122:: Death tests use fork(), which is unsafe particularly in a threaded context. For this test, Google Test detected 4 threads. See https://github.com/google/googletest/blob/master/googletest/docs/advanced.md#death-tests-and-threads for more explanation and suggested solutions, especially if this is the last message you see before your test times out.

Fixes: https://tracker.ceph.com/issues/43403
Signed-off-by: Sage Weil <sage@redhat.com>

qa: build v5.4 kernel

CentOS8 userspace cannot build v4.17 kernel anymore. See also [1].

[1] https://lkml.org/lkml/2019/4/23/517

Fixes: https://tracker.ceph.com/issues/43719
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #32761 from alimaredia/wip-rgw-verify-os-version-8.0

qa: pin rgw/verify to 8.0

doc/cephfs/client-auth: description and example are inconsistent

According to the path restriction example, the filesystem name
should be cephfs_a, not cephfs.  Converge on cephfs_a to avoid it
being confused with with the pool tag, which is always cephfs.

This was introduced in 160c4bfeb811 ("mon/AuthMonitor: Use new osd
auth caps for ceph fs authorize").

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

qa: pin rgw/verify to 8.0

Signed-off-by: Ali Maredia <amaredia@redhat.com>

mgr/dashboard: Update nodejs to v10.18.1

This will fix some problems that showed up after updating some npm packages.

Signed-off-by: Tiago Melo <tmelo@suse.com>

mgr/dashboard: Fix npm vulnerabilities

Add npm-force-resolutions to help fix npm vulnerabilities.

When you need to force a package to be updated,
add it to the resolutions lists in package.json and run "npm run fix:audit".

This will update package-lock.json and all next "npm ci" will contain the
specified version of the package.

Signed-off-by: Tiago Melo <tmelo@suse.com>

Merge PR #32748 into master

* refs/pull/32748/head:
doc: minor fix in mount.ceph

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #32751 from tchaikov/wip-readme-cmake

README.md: remove stale cmake prerequisite

Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-By: Casey Bodley <cbodley@redhat.com>

README.md: remove stale cmake prerequisite

it's stale. as we now require 3.10 after dropping the support of xenial.

and it's incomplete, we have a ton of build dependencies apart from
cmake.

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #32655 from jan--f/c-v-mock-has_bluestore_label

ceph-volume: fix device unittest, mock has_bluestore_label

Merge PR #32738 into master

* refs/pull/32738/head:
qa/standalone/misc/ok-to-stop: improve test
qa/standalone/ceph-helpers: add wait_for_peered

Reviewed-by: Josh Durgin <jdurgin@redhat.com>

Merge pull request #32740 from cbodley/wip-rgw-doc-s3-extension-url

doc/rgw: fix broken link to boto s3 extensions document

Reviewed-by: Yuval Lifshitz <yuvalif@yahoo.com>

Merge pull request #32742 from tchaikov/wip-crimson-thread-pool

crimson/thread: generalize Task so it works w/ func returns void

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

ceph-volume: fix device unittest, mock has_bluestore_label, lint

This also adds a test case for has_bluestore_label.

Fixes: https://tracker.ceph.com/issues/43611
Signed-off-by: Jan Fajerski <jfajerski@suse.com>