Repository navigation
Remove PMON and XCVRD delay on system boot for modules controlled by SW - #18295
vadymhlushko-mlnx wants to merge 2 commits into
Conversation
|
@vadymhlushko-mlnx is this cleanly cherry picked to 202311 or we need a dedicate PR? if such is required we should have the label set properly and relate to the new PR against 202311. |
I've already created a separate PR for 202311 - [202311][IM] Remove the PMON and XCVRD delay on system boot #18296 |
088b785 to
dd7317c
Compare
|
@prgeor could you please help to merge |
There was a problem hiding this comment.
@vadymhlushko-mlnx Why will Systemd not start the PMON early? Why not remove "After=Syncd" for your platform?
There was a problem hiding this comment.
@vadymhlushko-mlnx would you please check the comments?
There was a problem hiding this comment.
@prgeor XCVRD should be started after we create_switch and initialize all the ports, that is why we have a dependency for the syncd.
There was a problem hiding this comment.
@vadymhlushko-mlnx There is already a dependency between Xcvrd and Swss. Xcvrd waits for PortInitDone and untill portsyncd sets its after all host interfaces are created: https://github.com/sonic-net/sonic-swss/blob/19410232523283d99eaa211979cc1e58b6cb262f/portsyncd/portsyncd.cpp#L134
We are holding complete PMON init just for Xcvrd which doesn't seems right. Can we use PortInitDone as a synchronization?
There was a problem hiding this comment.
@prgeor
This PR goal is to start the XCVRD CMIS manager task as soon as possible to link UP the SW-controlled ports as quickly as possible during the fastboot.
The XCVRD CMISMangerTaks already using the PortInitDone sync mechanism.
There was a problem hiding this comment.
@vadymhlushko-mlnx synchronization via systemd service file is fine. Why do we need to start pmon from synch.sh?
There was a problem hiding this comment.
@vadymhlushko-mlnx let systemd do the synchronization between syncd and pmon service
There was a problem hiding this comment.
@prgeor Some services are started by shell scripts, some via systemd. Both approaches work in SONiC. Starting from shell script gives more flexibility
There was a problem hiding this comment.
@prgeor I've removed unused code from the syncd.sh as you requested
There was a problem hiding this comment.
@vadymhlushko-mlnx OK. is the PR still in draft or ready for review?
dd7317c to
0cd7f5f
Compare
0cd7f5f to
cf77741
Compare
cf77741 to
b33ccc8
Compare
b33ccc8 to
51b745e
Compare
|
202311 branch cherry-pick |
665dd46 to
eed6e6c
Compare
|
@prgeor , Can you please review as your comment was addressed ? |
prgeor
left a comment
There was a problem hiding this comment.
@vadymhlushko-mlnx please use systemd for synchronization between PMON and syncd
Signed-off-by: vadymhlushko-mlnx <vadymh@nvidia.com>
Signed-off-by: vadymhlushko-mlnx <vadymh@nvidia.com>
aab3be6 to
c400723
Compare
|
New PR will be opened by @stepanblyschak |
Why I did it
To support the
fastbootfeature alongsidemodule controlled by swfeature.Because if the feature is enabled we need the
XCVRDto start as soon as possible to identify if the port isSWorFWcontrolled and bring the port to the UP state.Work item tracking
How I did it
Remove the delays for the
PMONdocker andXCVRDdaemon on the SONiC image start.How to verify it
Run the sonic-mgmt/tests/platform_tests/test_advanced_reboot.py - test_fast_reboot
Which release branch to backport (provide reason below if selected)
Tested branch (Please provide the tested image version)
Description for the changelog
Link to config_db schema for YANG module changes
A picture of a cute animal (not mandatory but encouraged)