Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Merge upstream/release/2.6 into upstream/google/2.6 #15066

Merged
merged 19 commits into from
Sep 4, 2024
Merged

Merge upstream/release/2.6 into upstream/google/2.6 #15066

merged 19 commits into from
Sep 4, 2024

Conversation

jolivier23
Copy link
Contributor

Nasf-Fan and others added 19 commits August 28, 2024 10:46
@Nasf-Fan
DAOS-16347 object: partial coll-punch because of CPU yield - b26 (#15013
0099aa4 
)

If multiple shards for the same object reside on the same vos target
and CPU yield for DTX refresh during collectively punch such object,
then the local collective punch needs to restarted to avoid missing
to punch some object shards on current target.

Drop redundant daos_is_zero_dti() check in object RPC hander logic.

Signed-off-by: Fan Yong <fan.yong@intel.com>
@mchaarawi
DAOS-16371 il: do not return unsupported for fputs if ioil is loaded (#…
9ced2da 
…14989) (#15006)

Signed-off-by: Mohamad Chaarawi <mohamad.chaarawi@intel.com>
@kjacque
DAOS-16352 control: Handle cases with static ifaces (#14953) (#14991)
ffb4aaa 
Fabric interfaces defined statically in the daos_agent config
file are fundamentally different from those detected via hardware
scanning. They don't include information derived from the hardware
such as their true device class or fabric provider(s).

This patch adds some rigor to what is ignored regarding these
manually-defined interfaces.

- Ignore provider for statically-defined fabric interfaces, as we
  do not bother detecting it. They are assumed to be compatible with
  whatever provider the agent is using.
- Silence confusing "no interfaces requested" error from
  WaitFabricReady by not calling it if there are no interfaces to
  check.
- Remove some defunct logic related to detecting the provider in
  the multi-provider case. The agent may only use a single provider.

Signed-off-by: Kris Jacque <kris.jacque@intel.com>
@phender
DAOS-16334 test: (#15022) (#15032)
59c4f38 
Update the --control-metadata-path assignment for dmg config generate
command to avoid issues cleaning up the files generated.

Signed-off-by: Phil Henderson <phillip.henderson@intel.com>
@Nasf-Fan
DAOS-16385 dtx: fix DRAM leak during handle DTX collective RPC (#15009)
98fffb7 
The DRAM leak will happen on non-leader engines when collectively
punch object.

Signed-off-by: Fan Yong <fan.yong@intel.com>
@daltonbohning
DAOS-16463 test: remove get_host_log_files (#15030) (#15040)
a176bc5 
Remove get_host_log_files since it is no longer needed.
Also remove search_log which used get_host_log_files and is unused.

Signed-off-by: Dalton Bohning <dalton.bohning@intel.com>
@shimizukko
DAOS-16406 test: file_count_test_base.py - Don't obtain dfuse mount_d…
bd4cfd2 
…ir from test yaml (#14993) (#15023)

Remove the following from file_count_test_base.py
mount_dir = self.params.get("mount_dir", "/run/dfuse/*")

dfuse mount_dir used to be defined in the test yaml, but now the harness creates a
unique directory per test. mount_dir definitions were removed from the test yaml
in PR11453.

Affected tests are:
deployment/io_sys_admin.py
io/large_file_count.py
io/small_file_count.py

Signed-off-by: Makito Kano <makito.kano@intel.com>
@phender
DAOS-16366 test: Use agent/server config files from test directory (#…
cf7322a 
…14944) (#15033)

Use agent, control, and server config files placed in the common test directory instead of /etc/daos with a systemctl override configuration file.

Signed-off-by: Phil Henderson <phillip.henderson@intel.com>
@daltonbohning
DAOS-16462 test: remove server manager srv_timeout (#15029) (#15041)
761ec54 
remove server manager srv_timeout since it is unused

Signed-off-by: Dalton Bohning <dalton.bohning@intel.com>
@soumagne
DAOS-16315 mercury: update to 2.4.0rc5 (#15015) (#15039)
f4e75a7 
Signed-off-by: Jerome Soumagne <jerome.soumagne@intel.com>
@phender
DAOS-16484 test: Exclude local host in default interface selection (#…
f4ae244 
…15049) (#15051)

When including the local host in the default interface selection a
difference in ib0 speeds will cause the logic to select eth0 and then
the tcp provider.

Signed-off-by: Phil Henderson <phillip.henderson@intel.com>
@NiuYawei
DAOS-16465 vos: fix misused ip_hdl (#15034) (#15043)
1ccdef3 
vos_iter_param_t.ip_hdl is for passing dkey or akey tree open handle
in vos_iterate_key() case, don't always regard it as 'vos_pool' or
'vos_container'.

Signed-off-by: Niu Yawei <yawei.niu@intel.com>
@liw @yyyshi
DAOS-14544 rsvc: crt tree type usage fix (#14834) (#15005)
a4352e1 
- Change from flat to knomial tree for broadcasts

Signed-off-by: shiying <yshi1210@gmail.com>
Signed-off-by: Li Wei <wei.g.li@intel.com>
Co-authored-by: shiying <yshi1210@gmail.com>
@knard38
DAOS-14348 client: GC for pil4dfs dentry cache (#14996)
f12df7f 
Hotfix of miscellaneous issues:
- invalid type of atomic_flag
- invalid usage of DL_WARN macro

Signed-off-by: Cedric Koch-Hofer <cedric.koch-hofer@intel.com>
@wiliamhuang
DAOS-16365 client: intercept MPI_Init() to avoid nested call (#14992) (…
a0e7d52 
…#15047)

We observed deadlock in MPI applications on Aurora due to nested calls of zeInit() inside MPI_Init(). daos_init() is involved in such nested calls. This PR intercepts MPI_Init() and avoid running daos_init() inside MPI_Init().

Signed-off-by: Lei Huang <lei.huang@intel.com>
@mchaarawi
DAOS-16272 dfs: fix get_info returning incorrect oclass (#15048) (#15058
15aea1c 
)

If user creates a container without --file-oclass, the get_info call was
returning the default oclass of a directory on daos fs get-attr. Fix
that to properly use the enum types for default scenario.

Signed-off-by: Mohamad Chaarawi <mohamad.chaarawi@intel.com>
@kjacque
DAOS-16245 control: Fix dmg cont set-owner (#14945) (#15042)
fe3f42a 
- Fix bug where all CONT_PROP_SET RPCs were sent directly to
  ds_cont_op_hdlr instead of using the custom set_prop handler
  that detected server vs. client RPCs. ds_cont_op_hdlr now detects
  server RPCs and calls the appropriate handler.
- Add CONT_PROP_SET_BYLABEL RPC to allow for container label usage
  via dmg (without pool/cont handles).
- Update dmg cont set-owner command to accept positional arguments
  and allow either UUIDs or labels.

Signed-off-by: Kris Jacque <kris.jacque@intel.com>
@jolivier23
Merge remote-tracking branch 'comm/release/2.6' into jeffolivier/goog…
9294c1f 
…le/2.6
@jolivier23
Revert mercury v2.4.0rc5 for now
d1a0b75 
Required-githooks: true

Change-Id: Id091ec597b3ee5871d06e376106e88b588537608
Signed-off-by: Jeff Olivier <jeffolivier@google.com>
@github-actions GitHub Actions
Copy link

github-actions bot commented Sep 4, 2024

Errors are component not formatted correctly,Ticket number prefix incorrect,PR title is malformatted. See https://daosio.atlassian.net/wiki/spaces/DC/pages/11133911069/Commit+Comments,Unable to load ticket data
https://daosio.atlassian.net/browse/Merge

@karthjyojay karthjyojay self-requested a review September 4, 2024 18:43
@jolivier23 jolivier23 merged commit 687aedf into google/2.6 Sep 4, 2024
49 of 50 checks passed
@jolivier23 jolivier23 deleted the jeffolivier/google/2.6 branch September 4, 2024 23:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@karthjyojay karthjyojay karthjyojay approved these changes

Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.