[config] Exit with non-zero when qos reload fail#3710
[config] Exit with non-zero when qos reload fail#3710kperumalbfn merged 3 commits intosonic-net:masterfrom
Conversation
Fix issue when 'config qos reload' operation unsuccessful, but returns exit code 0, when operation unsuccessful, now it will return code 1 after executing all the commands. Signed-off-by: Jianyue Wu <jianyuew@nvidia.com>
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
Please coordinate this change with PR #3708 . |
|
@vivekrnv Please help review this change. |
config/main.py
Outdated
| # Traditional buffer manager do not remove buffer tables in any case, no need to wait. | ||
| timeout = 120 if device_metadata and device_metadata.get('buffer_model') == 'dynamic' else 0 | ||
| _wait_until_clear(["BUFFER_*_TABLE:*", "BUFFER_*_SET"], interval=0.5, timeout=timeout, verbose=verbose) | ||
| return _wait_until_clear(["BUFFER_*_TABLE:*", "BUFFER_*_SET"], interval=0.5, timeout=timeout, verbose=verbose) |
There was a problem hiding this comment.
Could you check for actual table/config returning error? Will be good for debugging
There was a problem hiding this comment.
Currently inside wait_until_clear() function, when enable --verbose mode, it will print out table and config error details.
$ config qos reload **--verbose
Some entries matching BUFFER_TABLE: still exist: BUFFER_PROFILE_TABLE:ingress_lossy_pg_zero_profile
Some entries matching BUFFER__TABLE: still exist: BUFFER_PROFILE_TABLE:ingress_lossy_pg_zero_profile**
Operation not completed successfully, please save and reload configuration.
Running command: /usr/local/bin/sonic-cfggen -d -t /usr/share/sonic/device/x86_64-mlnx_msn3800-r0/Mellanox-SN3800-D112C8/buffers_dynamic.json.j2,/tmp/cfg_buffer.json -t /usr/share/sonic/device/x86_64-mlnx_msn3800-r0/Mellanox-SN3800-D112C8/qos.json.j2,/tmp/cfg_qos.json -y /etc/sonic/sonic_version.yml
Running command: /usr/local/bin/sonic-cfggen -j /tmp/cfg_buffer.json -j /tmp/cfg_qos.json --write-to-db
| if buffer_model_updated: | ||
| print("Buffer calculation model updated, restarting swss is required to take effect") | ||
|
|
||
| if not status: |
There was a problem hiding this comment.
Add error log with additional details
There was a problem hiding this comment.
Previously in _wait_until_clear(), it has printed with this error, maybe user can refer to this one😊
- if not empty and timeout:
click.echo("Operation not completed successfully, please save and reload configuration.")
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Remove test_qos_wait_until_clear_not_empty_should_exit due to the special handling of timeout=0 indicating traditional buffer mode, making the test obsolete. Signed-off-by: Jianyue Wu <jianyuew@nvidia.com>
jianyuewu
force-pushed
the
config_qos_reload_fix
branch
from
62d37cf to
86a9205
Compare
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
vivekrnv added _wait_until_clear UT for special case timeout=0 in PR #3708, so remove in current PR to avoid conflict. |
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
jianyuewu
force-pushed
the
config_qos_reload_fix
branch
from
b02ce0a to
1724b7e
Compare
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
config qos reload --no-delay should return with 0, instead of 1. Signed-off-by: Jianyue Wu <jianyuew@nvidia.com>
jianyuewu
force-pushed
the
config_qos_reload_fix
branch
from
1724b7e to
a2f6d86
Compare
|
/azp run |
|
Azure Pipelines successfully started running 1 pipeline(s). |
jianyuewu
requested review from
congh-nvidia,
kperumalbfn,
stephenxs and
vivekrnv
|
Cherry-pick PR to 202411: #3727 |
mssonicbld
added
Included in 202411 Branch
and removed
Created PR to 202411 Branch
labels
Thank you. |
…fail (sonic-net#3710)" This reverts commit 1884b95.
[config] Exit with non-zero when qos reload fail
8 participants
Fix issue when 'config qos reload' operation unsuccessful, but returns exit code 0, when operation unsuccessful, now it will return code 1 after executing all the commands.
What I did
When config qos reload failed with error print "Operation not completed successfully", exit with code 1 finally, instead of exit with code 0.
How I did it
Check return value in function _wait_until_clear(), and after config qos reload command finish, check this flag, if it is none zero, then exit 1.
How to verify it
Test with command "config qos reload", will exit 1. Note that "config load_minigraph" is also verified, not impacted.
Previous command output (if the output of a command-line utility has changed)
# config qos reload Operation not completed successfully, please save and reload configuration. Running command: /usr/local/bin/sonic-cfggen -d -t /usr/share/sonic/device/x86_64-nvidia_sn5600-r0/Mellanox-SN5600-C256S1/buffers_dynamic.json.j2,/tmp/cfg_buffer.json -t /usr/share/sonic/device/x86_64-nvidia_sn5600-r0/Mellanox-SN5600-C256S1/qos.json.j2,/tmp/cfg_qos.json -y /etc/sonic/sonic_version.yml Running command: /usr/local/bin/sonic-cfggen -j /tmp/cfg_buffer.json -j /tmp/cfg_qos.json --write-to-db Buffer calculation model updated, restarting swss is required to take effectNew command output (if the output of a command-line utility has changed)