Collect asic info and store in CHASSIS_STATE_DB #175

ngoc-do · 2021-04-26T21:06:41Z

chassisd inside pmon queries asic information from platform modules and stores it into CHASSIS_STATE_DB of redis_chassis.

The reason we keep information in CHASSIS_STATE_DB because it's accessible by swss containers.

Asic information includes asic pci address, module name, and asic id in the module. For example, CHASSIS_STATE_DB will look like below:

127.0.0.1:6379[6]> HGETALL "CHASSIS_ASIC_TABLE|ASIC0"
1) "asic_pci_address"
2) "0000:09:00.0"
3) "module_name"
4) "FABRIC-CARD0"
5) "asic_id_in_module"
6) "0"
127.0.0.1:6379[6]> HGETALL "CHASSIS_ASIC_TABLE|ASIC1"
1) "asic_pci_address"
2) "0000:0b:00.0"
3) "module_name"
4) "FABRIC-CARD0"
5) "asic_id_in_module"
6) "1"
127.0.0.1:6379[6]> HGETALL "CHASSIS_ASIC_TABLE|ASIC2"
1) "asic_pci_address"
2) "0000:13:00.0"
3) "module_name"
4) "FABRIC-CARD1"
5) "asic_id_in_module"
6) "0"

This PR will merge only after sonic-net/sonic-platform-common#185 merges since this PR calls get_all_asics that is implemented in the other one.

sonic-chassisd/scripts/chassisd

judyjoseph · 2021-06-02T06:55:55Z

/azp run

azure-pipelines · 2021-06-02T06:56:04Z

Azure Pipelines successfully started running 1 pipeline(s).

judyjoseph · 2021-06-14T19:34:47Z

sonic-chassisd/scripts/chassisd

@@ -163,6 +170,9 @@ class ModuleUpdater(logger.Logger):
                               CHASSIS_MODULE_INFO_SLOT_FIELD,
                               CHASSIS_MODULE_INFO_OPERSTATUS_FIELD]

+        chassis_state_db = daemon_base.db_connect("CHASSIS_STATE_DB")


This new table is kept in CHASSIS_STATE_DB, unlike the other tables like MIDPLANE_TABLE etc that is kept in STATE_DB ?

See this comment in PR description that the info is kept in CHASSIS_STATE_DB because it's accessible by swss containers. Does the swss use this asic present info ?

There are a couple of places we can keep asic info like STATE_DB (redis instance) or CHASSIS_STATE_DB (redis chassis instance).

CHASSIS_STATE_DB is accessible from every where in swss like/usr/bin/swss.sh or /usr/bin/local/swss.sh - script creating swss containeranddocker_init.sh` inside swss container. Inside swss container, STATE_DB is not accessible (asic database is mapped, instead of database). So CHASSIS_STATE_DB is a safer choice.

anshuv-mfst · 2021-06-21T07:22:08Z

@judyjoseph - can you please merge asap please, needed for 202106 release

Add a generic function that returns a list of asics on module. Vendors will implement their own for their platform. The function could be used for collecting fabric asic info in each module and populating into CHASSIS_STATE_DB (sonic-net/sonic-platform-daemons#175). To simplify the change, all fabric cards in system should be identical in terms of number of asics in a card.

judyjoseph · 2021-06-22T00:12:57Z

/azp run

azure-pipelines · 2021-06-22T00:13:07Z

Azure Pipelines successfully started running 1 pipeline(s).

sonic-chassisd/tests/test_chassisd.py

Signed-off-by: ngocdo <ngocdo@arista.com>

mprabhu-nokia · 2021-06-23T16:28:51Z

@judyjoseph I don't think I have permissions to merge. Could you please help with that? I have approved.

Chassisd inside pmon queries asic information from platform modules and stores it into CHASSIS_STATE_DB of redis_chassis. The reason we keep information in CHASSIS_STATE_DB because it's accessible by swss containers. Asic information includes asic pci address, module name, and asic id in the module. For example, CHASSIS_STATE_DB will look like below: 127.0.0.1:6379[6]> HGETALL "CHASSIS_ASIC_TABLE|ASIC0" 1) "asic_pci_address" 2) "0000:09:00.0" 3) "module_name" 4) "FABRIC-CARD0" 5) "asic_id_in_module" 6) "0"

mprabhu-nokia · 2021-07-09T12:29:09Z

sonic-chassisd/scripts/chassisd

+        asics = list(self.asic_table.getKeys())
+        for asic in asics:
+           fvs = self.asic_table.get(asic)
+           if fvs[CHASSIS_MODULE_INFO_NAME_FIELD] in notOnlineModules:


@ngoc-do
Looks like there is a bug here? Are you seeing this as well? fvs is in the format below:

[True, (('asic_pci_address', 'n/a'), ('name', 'FABRIC-CARD6'), ('asic_id_in_module', '1'))]

However, if the code is changed to below, it works:

fvp = dict(fvs[1]) if fvp[CHASSIS_MODULE_INFO_NAME_FIELD] in notOnlineModules:

Hmm I don't see that. But if this happens, the unit test will fail, I think?

unit-tests use mock_swsscommon.py, where it uses a simple dictionary. So, unit-tests are passing. However, I am seeing the behavior I mentioned previously in the real hw environment. Are you testing on the near latest github? Will try and investigate. Will help if you can confirm/investigate as well.

Ah OK.

Yes, I tested on our chassis when committing the change. But it could be that there were new changes at the upstream, and I didn't have latest upstream in our local repos by the time I tested. I will double check, but I'm afraid that our local repos hasn't been updated at this moment. So if you see it needs to be fixed, free to do that.

OK, I found it happened on our latest update.

I have a fix in #203. Please review.

Add a generic function that returns a list of asics on module. Vendors will implement their own for their platform. The function could be used for collecting fabric asic info in each module and populating into CHASSIS_STATE_DB (sonic-net/sonic-platform-daemons#175). To simplify the change, all fabric cards in system should be identical in terms of number of asics in a card.

judyjoseph · 2021-09-02T23:27:44Z

This commit is already present in 202106 sonic-buildimage submodule ptr.

Fix a bug coming from PR #175 when asic is removed due to card getting offline (before it was online).

…t#175) Fix `get_status_led()` API definition to conform to the description. Signed-off-by: Volodymyr Boyko <volodymyrx.boiko@intel.com>

Asic PCI ID (PCI address) is collected by chassisd (inside pmon - sonic-net/sonic-platform-daemons#175) and saved in CHASSIS_STATE_DB (in redis_chassis). CHASSIS_STATE_DB is accessible by swss containers. At docker-init.sh (script is called after swss container is created and before anything that could run in swss like orchagent...), we wait until asic PCI ID of the corresponding asic is populated by chassisd. We then update asic_id in CONFIG_DB of asic's database. A system supporting dynamic asic PCI ID identification requires to have a file (empty) use_pci_id_chassis in its platform dir. When orchagent runs, it has correct asic PCI ID in its CONFIG_DB. Together with this PR: sonic-net/sonic-platform-daemons#175 sonic-net/sonic-platform-common#185 Signed-off-by: Maxime Lorrillere <mlorrillere@arista.com>

… CONFIG_DB (#9681) Asic PCI ID (PCI address) is collected by chassisd (inside pmon - sonic-net/sonic-platform-daemons#175) and saved in CHASSIS_STATE_DB (in redis_chassis). CHASSIS_STATE_DB is accessible by swss containers. At docker-init.sh (script is called after swss container is created and before anything that could run in swss like orchagent...), we wait until asic PCI ID of the corresponding asic is populated by chassisd. We then update asic_id in CONFIG_DB of asic's database. A system supporting dynamic asic PCI ID identification requires to have a file (empty) use_pci_id_chassis in its platform dir. When orchagent runs, it has correct asic PCI ID in its CONFIG_DB. Together with this PR: sonic-net/sonic-platform-daemons#175 sonic-net/sonic-platform-common#185 Signed-off-by: Maxime Lorrillere <mlorrillere@arista.com> Co-authored-by: Maxime Lorrillere <mlorrillere@arista.com>

mprabhu-nokia suggested changes Apr 27, 2021

View reviewed changes

sonic-chassisd/scripts/chassisd Show resolved Hide resolved

sonic-chassisd/scripts/chassisd Show resolved Hide resolved

mprabhu-nokia mentioned this pull request Apr 29, 2021

[systemd] ASIC status based service bringup on VOQ chassis sonic-net/sonic-buildimage#7477

Merged

4 tasks

ngoc-do changed the title ~~Collect asic info and store in database chassis' STATE_DB~~ Collect asic info and store in CHASSIS_STATE_DB May 19, 2021

ngoc-do mentioned this pull request May 19, 2021

Get asic PCI ID from CHASSIS_STATE_DB and update asic_id in CONFIG_DB sonic-net/sonic-buildimage#7649

Open

4 tasks

anshuv-mfst requested a review from judyjoseph June 2, 2021 16:35

ngoc-do mentioned this pull request Jun 4, 2021

Add a template function that returns list of asics on module sonic-net/sonic-platform-common#185

Merged

ngoc-do force-pushed the fabric branch from fcc9c90 to 2abc2ee Compare June 4, 2021 20:17

judyjoseph reviewed Jun 14, 2021

View reviewed changes

judyjoseph previously approved these changes Jun 14, 2021

View reviewed changes

judyjoseph reviewed Jun 22, 2021

View reviewed changes

sonic-chassisd/tests/test_chassisd.py Outdated Show resolved Hide resolved

judyjoseph added the Request for 202106 Branch label Jun 22, 2021

ngoc-do and others added 5 commits June 22, 2021 16:42

Collect asic info and store in database chassis' STATE_DB

4473710

Signed-off-by: ngocdo <ngocdo@arista.com>

Store asic PCI ID in CHASSIS_STATE_DB and delete table at deinit()

81f65ad

Signed-off-by: ngocdo <ngocdo@arista.com>

Add test

5430bbc

Fix an accidentally removed line

917c604

Change prefix test_ to verify_

ccd0ad8

ngoc-do dismissed judyjoseph’s stale review via ccd0ad8 June 22, 2021 23:45

ngoc-do force-pushed the fabric branch from b351592 to ccd0ad8 Compare June 22, 2021 23:45

judyjoseph approved these changes Jun 23, 2021

View reviewed changes

mprabhu-nokia approved these changes Jun 23, 2021

View reviewed changes

judyjoseph merged commit b2c6102 into sonic-net:master Jun 23, 2021

mprabhu-nokia reviewed Jul 9, 2021

View reviewed changes

ngoc-do mentioned this pull request Jul 14, 2021

Fix a bug at removing asics on not-online cards #203

Merged

judyjoseph removed the Request for 202106 Branch label Sep 2, 2021

judyjoseph pushed a commit that referenced this pull request Sep 14, 2021

Fix a bug in removing asics on not-online cards (#203)

dfe737a

Fix a bug coming from PR #175 when asic is removed due to card getting offline (before it was online).

judyjoseph pushed a commit that referenced this pull request Oct 14, 2021

Fix a bug in removing asics on not-online cards (#203)

25b32ec

Fix a bug coming from PR #175 when asic is removed due to card getting offline (before it was online).

mlorrillere mentioned this pull request Jan 5, 2022

Get asic PCI ID from CHASSIS_STATE_DB and update asic_id in CONFIG_DB sonic-net/sonic-buildimage#9681

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Collect asic info and store in CHASSIS_STATE_DB #175

Collect asic info and store in CHASSIS_STATE_DB #175

ngoc-do commented Apr 26, 2021 •

edited

Loading

judyjoseph commented Jun 2, 2021

azure-pipelines bot commented Jun 2, 2021

judyjoseph Jun 14, 2021

judyjoseph Jun 14, 2021

ngoc-do Jun 21, 2021

anshuv-mfst commented Jun 21, 2021

judyjoseph commented Jun 22, 2021

azure-pipelines bot commented Jun 22, 2021

mprabhu-nokia commented Jun 23, 2021

mprabhu-nokia Jul 9, 2021

ngoc-do Jul 9, 2021

mprabhu-nokia Jul 12, 2021

ngoc-do Jul 12, 2021

ngoc-do Jul 14, 2021

judyjoseph commented Sep 2, 2021 •

edited

Loading

Collect asic info and store in CHASSIS_STATE_DB #175

Collect asic info and store in CHASSIS_STATE_DB #175

Conversation

ngoc-do commented Apr 26, 2021 • edited Loading

judyjoseph commented Jun 2, 2021

azure-pipelines bot commented Jun 2, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

anshuv-mfst commented Jun 21, 2021

judyjoseph commented Jun 22, 2021

azure-pipelines bot commented Jun 22, 2021

mprabhu-nokia commented Jun 23, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

judyjoseph commented Sep 2, 2021 • edited Loading

ngoc-do commented Apr 26, 2021 •

edited

Loading

judyjoseph commented Sep 2, 2021 •

edited

Loading