In modular chassis, add CHASSIS_STATE_DB on control card #5624

mprabhu-nokia · 2020-10-14T08:00:08Z

HLD: sonic-net/SONiC#646

In modular chassis, add CHASSIS_STATE_DB on control card

Why I did it
Modular Chassis has control-cards, line-cards and fabric-cards along with other peripherals. Control-Card CHASSIS_STATE_DB will be the central DB to maintain any state information of cards that is accessible to control-card/
How I did it
Adding another DB on an existing REDIS instance running on port 6380.

lguohan · 2020-10-18T04:11:00Z

@BrynXu can you help to review?

lguohan · 2020-10-18T04:11:16Z

retest vsimage please

lguohan · 2020-10-18T04:11:39Z

can you resolve conflict?

qiluo-msft · 2020-10-18T20:10:49Z

There is no GLOBAL_STATE_DB mentioned in sonic-net/SONiC#646.
Why not use existing STATE_DB? You are free to create new table in that DB. #Closed

dockers/docker-database/database_config.json.j2

mprabhu-nokia · 2020-10-19T18:01:12Z

There is no GLOBAL_STATE_DB mentioned in Azure/SONiC#646.
Why not use existing STATE_DB? You are free to create new table in that DB.

Added additonal sections in the document with the temperature_info-schema for the table in CHASSIS_STATE_DB . This was discussed in Section 3.2.4.

shyam77git · 2020-10-21T05:55:45Z

There is no GLOBAL_STATE_DB mentioned in Azure/SONiC#646.
Why not use existing STATE_DB? You are free to create new table in that DB.

Added additonal sections in the document with the temperature_info-schema for the table in CHASSIS_STATE_DB . This was discussed in Section 3.2.4.

Enhancing CC/Supervisor's local redis-DB to add one more table catering to sensors contents? or
existing approach of keeping Global redis-DB separate from local 'redis-DB' on CC/Supervisor?

IMO, local redis-DB of CC is initialized and maintained by PMON's chassisD whereas Global redis-DB of CC is initialized and maintained by PMON's thermalctlD. Since these two are seperate daemons inside PMON container so their respective DB's to be kept independent and restart of one should not impact other

mprabhu-nokia · 2020-10-28T16:43:16Z

retest vsimage please

mprabhu-nokia · 2020-11-04T17:29:51Z

retest vsimage please

mprabhu-nokia · 2020-11-16T18:28:53Z

retest baseimage please

mprabhu-nokia · 2020-11-16T21:23:28Z

retest vsimage please

mprabhu-nokia · 2020-11-16T21:24:07Z

retest mellanox please

mprabhu-nokia · 2020-11-17T01:06:24Z

retest vsimage please

mprabhu-nokia · 2020-11-17T01:41:09Z

retest vsimage please

mprabhu-nokia · 2020-11-17T22:13:51Z

retest vsimage please

mprabhu-nokia · 2020-11-18T19:07:54Z

@BrynXu @judyjoseph

With this change, it fails in below test. Since you made changes to CHASSIS_APP_DB, are you aware why this could fail?
I thought there was a dependency on swss-common. Do we need to add this to "platform/vs/docker-sonic-vs/database_config.json" as well?

TASK [execute cli "config load_minigraph -y" to apply new minigraph] ***********

mprabhu-nokia · 2020-11-18T20:36:14Z

@BrynXu @judyjoseph

With this change, it fails in below test. Since you made changes to CHASSIS_APP_DB, are you aware why this could fail?
I thought there was a dependency on swss-common. Do we need to add this to "platform/vs/docker-sonic-vs/database_config.json" as well?
TASK [execute cli "config load_minigraph -y" to apply new minigraph] ***********

Looks like the error was because it was not able to find redis_chassis. Will update "files/scripts/update_chassisdb_config" to delete this from the config.

mprabhu-nokia · 2020-11-19T04:39:50Z

retest baseimage please

mprabhu-nokia · 2020-11-19T07:20:51Z

retest vs please

mprabhu-nokia · 2020-11-19T07:21:15Z

retest vsimage please

mprabhu-nokia · 2020-11-19T17:52:59Z

@BrynXu @qiluo-msft please review

mprabhu-nokia · 2020-11-19T20:15:25Z

retest baseimage please

On a modular chassis, add a CHASSIS_STATE_DB where all state information from line-cards can be persisted.

mprabhu-nokia · 2020-11-20T09:21:22Z

retest vsimage please

mprabhu-nokia · 2020-11-20T17:37:22Z

retest vsimage please

mprabhu-nokia · 2020-11-21T03:31:30Z

@qiluo-msft @judyjoseph @jleveque could one of you please review/approve this?

judyjoseph · 2020-11-23T21:48:10Z

Can you check this PR ..#5283 if you need references of CHASSIS_STATE_DB in vs_platform as well ?

mprabhu-nokia · 2020-11-23T23:39:49Z

Can you check this PR ..#5283 if you need references of CHASSIS_STATE_DB in vs_platform as well ?

@judyjoseph since the CHASSIS_STATE_DB provider/consumers are pmon processes, there was no need to add references in vs platform.

qiluo-msft

The only concern is occupying new Redis database index. I prefer other reviewers familiar with this topic to approve.

mprabhu-nokia · 2020-12-05T01:44:01Z

retest vs please

judyjoseph · 2020-12-08T18:46:34Z

Discussed with @Qi on this, the suggestion he was making is not to define a new DB for chassis APP/STATE DB .. instead use the existing STATE_DB/APP_DB , and created tables in them with CHASSIS_ prefix in them.
@BrynXu was there any discussion why CHASSIS_APP_DB was added as a new DB with redis instance.

minionatwork · 2020-12-09T19:19:33Z

Discussed with @Qi on this, the suggestion he was making is not to define a new DB for chassis APP/STATE DB .. instead use the existing STATE_DB/APP_DB , and created tables in them with CHASSIS_ prefix in them.
@BrynXu was there any discussion why CHASSIS_APP_DB was added as a new DB with redis instance.

We introduced new DB because this new DB is hosted in supervisor card and reachable from all linecards. Existing ones aren't reachable to all linecards sonic instances and we would want to them to be isolated than shared.

CHASSIS_STATE_DB is added as part of this PR to keep pmon inline with using STATE_DB instead of APP_DB.

judyjoseph · 2020-12-09T20:00:29Z

Discussed with @Qi on this, the suggestion he was making is not to define a new DB for chassis APP/STATE DB .. instead use the existing STATE_DB/APP_DB , and created tables in them with CHASSIS_ prefix in them.
@BrynXu was there any discussion why CHASSIS_APP_DB was added as a new DB with redis instance.

We introduced new DB because this new DB is hosted in supervisor card and reachable from all linecards. Existing ones aren't reachable to all linecards sonic instances and we would want to them to be isolated than shared.

CHASSIS_STATE_DB is added as part of this PR to keep pmon inline with using STATE_DB instead of APP_DB.

Thanks ! I am fine with that then. A question though is Can we start the id from 0, 1 for this instance "redis_chassis", instead of continuing the numbering as 12, 13 with the original "redis" instance. Did you see any issues in your tests ?

BrynXu · 2020-12-09T20:13:34Z

we have an open issue,Duplicated dbid in redis multi-instances causes error <#5351>, to prevent us using duplicated dbid even they are in different redis instance.

…

On Wed, Dec 9, 2020 at 12:00 PM judyjoseph ***@***.***> wrote: Discussed with @Qi <https://github.com/Qi> on this, the suggestion he was making is not to define a new DB for chassis APP/STATE DB .. instead use the existing STATE_DB/APP_DB , and created tables in them with CHASSIS_ prefix in them. @BrynXu <https://github.com/BrynXu> was there any discussion why CHASSIS_APP_DB was added as a new DB with redis instance. We introduced new DB because this new DB is hosted in supervisor card and reachable from all linecards. Existing ones aren't reachable to all linecards sonic instances and we would want to them to be isolated than shared. CHASSIS_STATE_DB is added as part of this PR to keep pmon inline with using STATE_DB instead of APP_DB. Thanks ! I am fine with that then. A question though is Can we start the id from 0, 1 for this instance "redis_chassis", instead of continuing the numbering as 12, 13 with the original "redis" instance. Did you see any issues in your tests ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#5624 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADTR7VJEBDEON6OL3Y6BOX3ST7JOZANCNFSM4SQIE7IA> .

minionatwork · 2020-12-09T22:52:15Z

Discussed with @Qi on this, the suggestion he was making is not to define a new DB for chassis APP/STATE DB .. instead use the existing STATE_DB/APP_DB , and created tables in them with CHASSIS_ prefix in them.
@BrynXu was there any discussion why CHASSIS_APP_DB was added as a new DB with redis instance.

We introduced new DB because this new DB is hosted in supervisor card and reachable from all linecards. Existing ones aren't reachable to all linecards sonic instances and we would want to them to be isolated than shared.
CHASSIS_STATE_DB is added as part of this PR to keep pmon inline with using STATE_DB instead of APP_DB.

Thanks ! I am fine with that then. A question though is Can we start the id from 0, 1 for this instance "redis_chassis", instead of continuing the numbering as 12, 13 with the original "redis" instance. Did you see any issues in your tests ?

As BrynXu mentioned, there is some limitation in reusing ID. thats why it has to be unique across all DB instances even if its different redis server.

judyjoseph · 2020-12-09T22:58:45Z

Discussed with @Qi on this, the suggestion he was making is not to define a new DB for chassis APP/STATE DB .. instead use the existing STATE_DB/APP_DB , and created tables in them with CHASSIS_ prefix in them.
@BrynXu was there any discussion why CHASSIS_APP_DB was added as a new DB with redis instance.

We introduced new DB because this new DB is hosted in supervisor card and reachable from all linecards. Existing ones aren't reachable to all linecards sonic instances and we would want to them to be isolated than shared.
CHASSIS_STATE_DB is added as part of this PR to keep pmon inline with using STATE_DB instead of APP_DB.

Thanks ! I am fine with that then. A question though is Can we start the id from 0, 1 for this instance "redis_chassis", instead of continuing the numbering as 12, 13 with the original "redis" instance. Did you see any issues in your tests ?

As BrynXu mentioned, there is some limitation in reusing ID. thats why it has to be unique across all DB instances even if its different redis server.

Ok, we will revisit this along this issue resolution, #5351.

lguohan requested a review from qiluo-msft October 18, 2020 05:33

BrynXu reviewed Oct 18, 2020

View reviewed changes

dockers/docker-database/database_config.json.j2 Outdated Show resolved Hide resolved

dockers/docker-database/database_config.json.j2 Outdated Show resolved Hide resolved

BrynXu reviewed Oct 19, 2020

View reviewed changes

dockers/docker-database/database_config.json.j2 Outdated Show resolved Hide resolved

mprabhu-nokia closed this Oct 19, 2020

mprabhu-nokia reopened this Oct 19, 2020

mprabhu-nokia changed the title ~~In modular chassis, add GLOBAL_STATE_DB on control card~~ In modular chassis, add CHASSIS_STATE_DB on control card Nov 4, 2020

mprabhu-nokia force-pushed the nokia-thermal-pr branch from 2a3b9d0 to a414204 Compare November 16, 2020 08:13

mprabhu-nokia force-pushed the nokia-thermal-pr branch from c5af325 to e486786 Compare November 19, 2020 20:05

Introduce CHASSIS_STATE_DB for modular chassis

d9dea66

On a modular chassis, add a CHASSIS_STATE_DB where all state information from line-cards can be persisted.

Add CHASSIS_STATE_DB in update_chassisdb_config

2f6edec

mprabhu-nokia force-pushed the nokia-thermal-pr branch from e486786 to 2f6edec Compare November 20, 2020 03:12

qiluo-msft reviewed Nov 25, 2020

View reviewed changes

Merge branch 'master' into nokia-thermal-pr

69ea8f6

judyjoseph approved these changes Dec 9, 2020

View reviewed changes

judyjoseph merged commit 41012f7 into sonic-net:master Dec 16, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

In modular chassis, add CHASSIS_STATE_DB on control card #5624

In modular chassis, add CHASSIS_STATE_DB on control card #5624

mprabhu-nokia commented Oct 14, 2020 •

edited

Loading

lguohan commented Oct 18, 2020

lguohan commented Oct 18, 2020

lguohan commented Oct 18, 2020

qiluo-msft commented Oct 18, 2020 •

edited

Loading

mprabhu-nokia commented Oct 19, 2020

shyam77git commented Oct 21, 2020

mprabhu-nokia commented Oct 28, 2020

mprabhu-nokia commented Nov 4, 2020

mprabhu-nokia commented Nov 16, 2020

mprabhu-nokia commented Nov 16, 2020

mprabhu-nokia commented Nov 16, 2020

mprabhu-nokia commented Nov 17, 2020

mprabhu-nokia commented Nov 17, 2020

mprabhu-nokia commented Nov 17, 2020

mprabhu-nokia commented Nov 18, 2020

mprabhu-nokia commented Nov 18, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 20, 2020

mprabhu-nokia commented Nov 20, 2020

mprabhu-nokia commented Nov 21, 2020

judyjoseph commented Nov 23, 2020

mprabhu-nokia commented Nov 23, 2020

qiluo-msft left a comment

mprabhu-nokia commented Dec 5, 2020

judyjoseph commented Dec 8, 2020

minionatwork commented Dec 9, 2020

judyjoseph commented Dec 9, 2020

BrynXu commented Dec 9, 2020 via email

minionatwork commented Dec 9, 2020

judyjoseph commented Dec 9, 2020 •

edited

Loading

In modular chassis, add CHASSIS_STATE_DB on control card #5624

In modular chassis, add CHASSIS_STATE_DB on control card #5624

Conversation

mprabhu-nokia commented Oct 14, 2020 • edited Loading

lguohan commented Oct 18, 2020

lguohan commented Oct 18, 2020

lguohan commented Oct 18, 2020

qiluo-msft commented Oct 18, 2020 • edited Loading

mprabhu-nokia commented Oct 19, 2020

shyam77git commented Oct 21, 2020

mprabhu-nokia commented Oct 28, 2020

mprabhu-nokia commented Nov 4, 2020

mprabhu-nokia commented Nov 16, 2020

mprabhu-nokia commented Nov 16, 2020

mprabhu-nokia commented Nov 16, 2020

mprabhu-nokia commented Nov 17, 2020

mprabhu-nokia commented Nov 17, 2020

mprabhu-nokia commented Nov 17, 2020

mprabhu-nokia commented Nov 18, 2020

mprabhu-nokia commented Nov 18, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 19, 2020

mprabhu-nokia commented Nov 20, 2020

mprabhu-nokia commented Nov 20, 2020

mprabhu-nokia commented Nov 21, 2020

judyjoseph commented Nov 23, 2020

mprabhu-nokia commented Nov 23, 2020

qiluo-msft left a comment

Choose a reason for hiding this comment

mprabhu-nokia commented Dec 5, 2020

judyjoseph commented Dec 8, 2020

minionatwork commented Dec 9, 2020

judyjoseph commented Dec 9, 2020

BrynXu commented Dec 9, 2020 via email

minionatwork commented Dec 9, 2020

judyjoseph commented Dec 9, 2020 • edited Loading

mprabhu-nokia commented Oct 14, 2020 •

edited

Loading

qiluo-msft commented Oct 18, 2020 •

edited

Loading

judyjoseph commented Dec 9, 2020 •

edited

Loading