Skip to content

Adding a .travis.yml file to trigger CI builds. - #1

Closed
rajachan wants to merge 1 commit into
daos-stack:masterfrom
rajachan:master
Closed

rajachan wants to merge 1 commit into
daos-stack:masterfrom
rajachan:master

Conversation

@rajachan

Copy link
Copy Markdown

Added a .travis.yml file for automatic CI builds in Travis. Currently just kicking off scons --build-deps=yes, which fails due to a CCI build issue. If you guys already have a way to build DAOS working around this issue, we can update the yml to reflect that.

You can see a sample build in my fork's TravisCI.

@ashleypittman

Copy link
Copy Markdown
Contributor

Our master builds are blocked on the CCI build issue as well so we're currently working from a blessed build that predates that bug. Hopefully it'll be resolved soon, and then we'll have to update from some Mercury API changes related to how checksums are calculated.

@johannlombardi

Copy link
Copy Markdown
Contributor

Thanks for the patch Raghu. We currently use gerrit for patch review. I have pushed your patch here: https://review.whamcloud.com/#/c/27918/
Please let me know if you are ok with the signed-off-by and I will merge the patch. Thanks!

@rajachan

rajachan commented Jul 3, 2017

Copy link
Copy Markdown
Author

Sure, I am okay with the signed-off-by.

@johannlombardi

Copy link
Copy Markdown
Contributor

Landed. Thanks.

@daos-jenkins

Copy link
Copy Markdown
Collaborator

test build

liw added a commit to liw/daos that referenced this pull request Jul 4, 2018
The following assertion failure was seen during the REBUILD test:

  rdb_appendentries_handler() 428d0eb9[6]: handling raft ae from rank 1
  rdb_raft_cb_debug() 428d0eb9[6]: recvd appendentries t:3 ci:227 lc:225
  pli:226 plt:2 daos-stack#1: rank=1
  rdb_lc_discard() lc=7f24ec986b30 low=227 high=227
  rdb_raft_cb_log_pop() 428d0eb9[6]: deleted entry 227: term=2 type=0
  buf=(nil) len=0
  rdb_raft_cb_log_offer() 227 == 228

The rdb_raft_cb_log_pop call forgot to update the log tail index,
causing the assertion in rdb_raft_cb_log_offer to fail.

Change-Id: I4c8c8d95263b7562c35819b851ce35f11a744034
Signed-off-by: Li Wei <wei.g.li@intel.com>
Reviewed-on: https://review.hpdd.intel.com/32783
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Omkar Kulkarni <omkar.kulkarni@intel.com>
Reviewed-by: Xuezhao Liu <xuezhao.liu@intel.com>
Reviewed-by: Liang Zhen <liang.zhen@intel.com>
mallove79 added a commit that referenced this pull request Aug 16, 2020
Clean up (hopefully unneccessary) cart.log printers. Put cart.log's into
default DAOS logs dir from the very beginning.

Competing approaches:
  1. Find where cart.log's are currently, and move them to artifacts
     dir.
  2. Put cart.log's into artifacts dir from the start.

Up to now, this branch has used approach #1. #2 appears to be the far
simpler approach; so KISS, reduce complexity -- try #2.

Skip-build-leap15-rpm: true
Skip-build-ubuntu-clang: true
Skip-build-leap15-icc: true
Skip-coverity-test: true
Skip-checkpatch: true
Quick-build: true
Skip-run_test: true
Test-tag: all_cart
Skip-func-test-leap15: true
Skip-func-hw-test-small: true
Skip-func-hw-test-medium: true
Skip-func-hw-test-large: true

Signed-off-by: Ethan Mallove <ethanx.a.mallove@intel.com>
tanabarr added a commit that referenced this pull request Sep 22, 2020
Signed-off-by: Tom Nabarro <tom.nabarro@intel.com>
dpquigl added a commit that referenced this pull request Nov 21, 2020
After peer review the following modifications were requested.

* Remove cipher suite AES128-GCM-SHA256 and just offer AES256-GCM-SHA384 as
default.

* Reduce certificates to one year lifetime (currently three years)

* Replace PKCS #1 v1.2 with PKCS #1 v2.1/2 if possible. The golang core library
is being used and may not support it yet. Team to verify if it does and, if so,
will try to update.

* Add to documentation recommending to customer to use the latest version of
OpenSSL >= 1.1.1h. For this project, the customer is responsible for generating
keys/certs.

* Reduce key sizes to 3072 based on expected lifetime of this product

Signed-off-by: David Quigley <david.quigley@intel.com>
mjmac pushed a commit that referenced this pull request Nov 24, 2020
After peer review the following modifications were requested:
  * Remove cipher suite AES128-GCM-SHA256 and just offer
    AES256-GCM-SHA384 as default.
  * Reduce certificates to one year lifetime (currently three years)
  * Replace PKCS #1 v1.2 with PKCS #1 v2.1/2 if possible. The golang core
    library is being used and may not support it yet. Team to verify if it
    does and, if so, will try to update.
  * Add to documentation recommending to customer to use the latest version of
    OpenSSL >= 1.1.1h. For this project, the customer is responsible for
    generating keys/certs.
  * Reduce key sizes to 3072 based on expected lifetime of this product

Signed-off-by: David Quigley <david.quigley@intel.com>
jolivier23 pushed a commit that referenced this pull request Jan 27, 2021
For these directoriesi and their sub_dirs:
daos/src/vos
daos/src/vea
daos/src/utils
daos/src/tests
daos/src/security

Signed-off-by: Hua Kuang <hua.kuang@intel.com>
ravalsam pushed a commit that referenced this pull request Feb 18, 2021
Fixed Coverity issue in test code.
313028(#2 of 2) and 313033(#1 of 1).

Signed-off-by: Samir Raval <samir.raval@intel.com>
mallove79 added a commit that referenced this pull request Feb 24, 2021
Fix coverity STRING_OVERFLOW issue:
  CID 318951 (#1 of 1): Copy into fixed size buffer (STRING_OVERFLOW) 39.
  fixed_size_dest: You might overrun the 8-character fixed-size string exp_status
  by copying cursorCopy + groupArray[g].rm_so without checking the length.

Skip-build-leap15-rpm: true
Skip-func-test-leap15: true
Skip-build-ubuntu-clang: true
Skip-build-leap15-icc: true
Skip-coverity-test: true
Quick-build: true
Test-tag: cart
Skip-unit-tests: true
Skip-nlt: true
Skip-unit-test: true
Skip-unit-test-memcheck: true
Skip-coverity-test: true
Skip-func-hw-test-small: true
Skip-func-hw-test-medium: true
Skip-func-hw-test-large: true

Signed-off-by: Ethan Mallove <ethanx.a.mallove@intel.com>
sylviachanoiyee pushed a commit that referenced this pull request Feb 26, 2021
Fixed Coverity issue in test code.
313028(#2 of 2) and 313033(#1 of 1).

Signed-off-by: Samir Raval <samir.raval@intel.com>
sylviachanoiyee pushed a commit that referenced this pull request Mar 11, 2021
Fix coverity issue:
  CID 318952 (#3 of 3): Explicit null dereferenced (FORWARD_NULL)
  15. var_deref_op: Dereferencing null pointer rank_list.

alloc_fn: Storage is returned from allocation function malloc.
  2. var_assign: Assigning: swim_seq = storage returned from malloc(1024UL).
     CID 318948 (#2 of 2): Resource leak (RESOURCE_LEAK)
  4. overwrite_var: Overwriting swim_seq in swim_seq = "" leaks the storage that
     swim_seq points to.

Fix coverity STRING_OVERFLOW issue:
  CID 318951 (#1 of 1): Copy into fixed size buffer (STRING_OVERFLOW) 39.
  fixed_size_dest: You might overrun the 8-character fixed-size string exp_status
  by copying cursorCopy + groupArray[g].rm_so without checking the length.

Signed-off-by: Ethan Mallove <ethanx.a.mallove@intel.com>
osalyk pushed a commit to osalyk/daos that referenced this pull request Oct 12, 2023
vos: local_transaction test with pre-step
dinghwah added a commit that referenced this pull request Jan 4, 2024
@PR-414#1

Skip-fnbullseye: false
Skip-bullseye: false
Skip-python-bandit: true
Skip-build-EL9-rpm: true
Allow-unstable-test: true
Doc-only: false
Test-nvme: auto_md_on_ssd
Test-tag: pr daily_regression full_regression
Required-githooks: true
Signed-off-by: Ding Ho ding-hwa.ho@intel.com
jolivier23 added a commit that referenced this pull request Jul 8, 2024
Signed-off-by: Jeff Olivier <jeffolivier@google.com>
liw added a commit that referenced this pull request Jul 14, 2026
To avoid leaking container children during reintegration, perform
container recovery in all reintegration modes. Not addressed by this
simple change is the extra delay for a POOL_REINT request.

Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Fix an ilog_modify log message that can't be controled by DD_MASK,
making it unnecessarily harder to turn on debug messages.

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit that referenced this pull request Jul 14, 2026
To avoid leaking container children during reintegration, perform
container recovery in all reintegration modes. Not addressed by this
simple change is the extra delay for a POOL_REINT request.

Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Fix an ilog_modify log message that can't be controled by DD_MASK,
making it unnecessarily harder to turn on debug messages.

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit to liw/daos that referenced this pull request Jul 15, 2026
To avoid leaking container children during reintegration, perform
container recovery in all reintegration modes. Not addressed by this
simple change is the extra delay for a POOL_REINT request.

Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  daos-stack#1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  daos-stack#2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  daos-stack#3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Fix an ilog_modify log message that can't be controled by DD_MASK,
making it unnecessarily harder to turn on debug messages.

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit that referenced this pull request Jul 16, 2026
To avoid leaking container children during reintegration, perform
container recovery in all reintegration modes. Not addressed by this
simple change is the extra delay for a POOL_REINT request.

Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Fix an ilog_modify log message that can't be controled by DD_MASK,
making it unnecessarily harder to turn on debug messages.

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit that referenced this pull request Jul 31, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit that referenced this pull request Jul 31, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit to liw/daos that referenced this pull request Aug 3, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  daos-stack#1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  daos-stack#2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  daos-stack#3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit to liw/daos that referenced this pull request Aug 3, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  daos-stack#1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  daos-stack#2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  daos-stack#3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit to liw/daos that referenced this pull request Aug 3, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  daos-stack#1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  daos-stack#2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  daos-stack#3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit to liw/daos that referenced this pull request Aug 4, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  daos-stack#1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  daos-stack#2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  daos-stack#3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit to liw/daos that referenced this pull request Aug 4, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  daos-stack#1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  daos-stack#2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  daos-stack#3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit to liw/daos that referenced this pull request Aug 4, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  daos-stack#1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  daos-stack#2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  daos-stack#3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit that referenced this pull request Aug 4, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
liw added a commit that referenced this pull request Aug 4, 2026
Fix segfault

  #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
  #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
  #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
  #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Features: rebuild
Signed-off-by: Li Wei <liwei@hpe.com>
daltonbohning pushed a commit that referenced this pull request Aug 31, 2026
To avoid leaking container children during reintegration, perform
container recovery in all reintegration modes. Not addressed by this
simple change is the extra delay for a POOL_REINT request.

Fix the backoff-less retrying of -DER_AGAINs from container recovery. It
has been seen to generate 11 million log messages within 2 minutes.

Fix an ilog_modify log message that can't be controled by DD_MASK,
making it unnecessarily harder to turn on debug messages.

Fix segfault

    #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
    #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
    #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
    #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Signed-off-by: Li Wei <liwei@hpe.com>
gnailzenh pushed a commit that referenced this pull request Sep 23, 2026
To avoid leaking container children during reintegration, perform
container recovery in all reintegration modes. Not addressed by this
simple change is the extra delay for a POOL_REINT request.

Fix the backoff-less retrying of -DER_AGAINs from container recovery. It
has been seen to generate 11 million log messages within 2 minutes.

Fix an ilog_modify log message that can't be controled by DD_MASK,
making it unnecessarily harder to turn on debug messages.

Fix segfault

    #0  crt_hg_bulk_transfer at src/cart/crt_hg.c:2253
    #1  crt_bulk_transfer at src/cart/crt_bulk.c:306
    #2  ds_pool_recov_cont_handler at src/pool/srv_target.c:3203
    #3  crt_handle_rpc at src/cart/crt_rpc.c:1756

that happens when reintegrating a target on the PS leader, likely caused
by NULL crt_rpc_priv.crp_hg_hdl. The approach is to use the existing
bulk mechanism offered by CaRT for CoRPCs. (The CaRT bug with local bulk
transfers is not addressed by this fix.)

Signed-off-by: Li Wei <liwei@hpe.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Development

Successfully merging this pull request may close these issues.

4 participants