gbsyncd fails to do PHY device cleanup OR graceful deinit during docker restart and config reload.

[jaganbal-a]

Problem description:

1. Gbsyncd fails to do PHY device cleanup OR graceful deinit during docker restart and config reload.

2. Restarting ‘syncd0’ docker, invokes /usr/bin/syncd_request_shutdown from syncd.sh which gbsyncd0 also receives the request and does the PHY cleanup, Which is incorrect.

Findings:

1. Since “stopplatform1” is not implemented in files/scripts/gbsyncd.sh, gbsyncd is not getting notification for graceful shutdown.
2. /usr/bin/docker exec -i /usr/bin/syncd_request_shutdown –cold results in shutdown request on syncd process in both gbsyncd/syncd dockers, which is triggered from syncd.sh when syncd docker goes for shutdown/restart.

So how to make syncnd_request_shutdown command to notify only the respective docker syncd process ?

root@sonic:/home/cisco# /usr/bin/docker exec -i syncd0 /usr/bin/syncd_request_shutdown --cold
requested COLD shutdown

Sep 20 16:00:52.631030 sonic NOTICE syncd0#syncd_request_shutdown: :- loadFromFile: no context config specified, will load default context config
Sep 20 16:00:52.631030 sonic NOTICE syncd0#syncd_request_shutdown: :- insert: added switch: idx 0, hwinfo ''
Sep 20 16:00:52.631227 sonic NOTICE syncd0#syncd_request_shutdown: :- send: requested COLD shutdown
Sep 20 16:00:52.631447 sonic NOTICE syncd0#syncd: :- run: is asic queue empty: 1
Sep 20 16:00:52.631447 sonic NOTICE syncd0#syncd: :- run: drained queue
Sep 20 16:00:52.631467 sonic NOTICE gbsyncd0#GBSAI[14]: :- run: is asic queue empty: 1
Sep 20 16:00:52.631499 sonic NOTICE gbsyncd0#GBSAI[14]: :- run: drained queue
Sep 20 16:00:52.631499 sonic NOTICE syncd0#syncd: :- handleRestartQuery: received COLD switch shutdown event
Sep 20 16:00:52.631541 sonic NOTICE gbsyncd0#GBSAI[14]: :- handleRestartQuery: received COLD switch shutdown event
Sep 20 16:00:52.632546 sonic NOTICE gbsyncd0#GBSAI[14]: :- removeAllSwitches: Removing all switches

 ---------------------------------------------------
 BUG REPORT INFORMATION
 ---------------------------------------------------
 Use the commands below to provide key information from your environment:
 You do NOT have to include this information if this is a FEATURE REQUEST

-->

Description

1. Gbsyncd fails to do PHY device cleanup OR graceful deinit during docker restart and config reload.
2. Restarting ‘syncd0’ docker, invokes /usr/bin/syncd_request_shutdown from syncd.sh which gbsyncd0 also receives the request and does the PHY cleanup, Which is incorrect.

Steps to reproduce the issue:

1. gbsyncd restart
2. config reload
3. execute the command /usr/bin/docker exec -i syncd0 /usr/bin/syncd_request_shutdown --cold

Describe the results you received:

1. With config reload, gbsyncd never receive notification to gracefully deinit the PHY. this is due to shutdownplatform1 is not implemented in gyncd.sh.
   Since the gbsyncd docker get terminated ~12 sec before syncd docker, the syncd_shutsown_request issued in syncd docker do not reach gbsyncd as it is already terminated.

2. syncd docker restart makes the gbsyncd to cleanup PHY .

Describe the results you expected:

gbsyncd should be notified of graceful de-init/shutdown during config reload.
syncd docker restart should only cleanup the device controlled by the docker and do not send notification to gbsyncd.

Output of show version:

Output of show techsupport:

[judyjoseph]

@jaganbal-a Could you share more info on which platform, which branch and build.

[jaganbal-a]

@jaganbal-a Could you share more info on which platform, which branch and build.

Platform: Multi-ASIC Platform with PHY device: I see this issue with Cisco 88-LC36-FH-MO line card.
Branch : azure/cisco/msft/202205, which is in sync with 202205 branch commit 7c68be0.
Build: It is cisco internal build.

[judyjoseph]

@abdosi f.y.i