SG350X-24 fans reach level 5 after stacking in ring configuration

[Emmo]

Hi, i configurated half a dozen of sites with SG350X switches connected in stack mode.

Every time i close the ring (topology change from chain to ring) the system immediately notify fan acceleration and every switch reach level 5 fan speed:

%STCK SYSL-I-UNITTRP: UNIT ID 2,Trap:%Environment-I-FANS-SPEED-CHNG: FAN'S speed level - 4 changed to level - 5.
%STCK SYSL-I-UNITTRP: UNIT ID 3,Trap:%Environment-I-FANS-SPEED-CHNG: FAN'S speed level - 4 changed to level - 5.
%STCK SYSL-I-UNITTRP: UNIT ID 3,Trap:%Environment-I-FANS-SPEED-CHNG: FAN'S speed level - 3 changed to level - 4.
%STCK SYSL-I-UNITTRP: UNIT ID 2,Trap:%Environment-I-FANS-SPEED-CHNG: FAN'S speed level - 3 changed to level - 4.
%STCK SYSL-I-UNITTRP: UNIT ID 2,Trap:%Environment-I-FANS-SPEED-CHNG: FAN'S speed level - 2 changed to level - 3.

The systems are in native stacking, on Gigabit Ethernet ports XG1 and XG2, firmware 2.5.5.47

Is that an expected behaviour?

It seems weird to me, because the CPU load remains very low.

Thank you in advance for every suggestion.

[balaji.bandi]

seems to be bug still persists in new version, if you have contract raise a SMB Tac case ?

https://community.cisco.com/t5/switching/sg350x-24-switch-overheating-shuts-down-ports-under-low-stress/td-p/3862361

[Emmo]

Thanks Balaji, i had read the post you linked but i hadn't figured out that the problem was due to the same reason.

[Michal Bruncko · Accepted solution]

https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvq41525/?rfs=iqvred

Last Modified: Oct 12,2020

Status: Terminated

Severity: 3 Moderate

Workaround: downgrade firmware to pre-2.5.x.x version

[Michal Bruncko]

Just to sum-up the behavour:

Setup:

• switch stack comprised of two SG350X-48 switches (native stack)
• interconnected with two direct stack links over 2x10G interfaces (Te1 and Te2 on both switches) (in our setup using copper cables) (Stack topology: Chain)
• switch software version: 2.5.5.47

Once the second stack link is connected and activated:

• Temperature sensor no.5 on PHY location (new sensor in 2.5.x.x) raises by 15 degrees of C
• FANs speed raises up to 100% within next 5 minutes (LVL 5)
• CPU utilization keep below 10% (no visible increase)

Immediately after disconnection of one of stack links:

• the FAN speed slows back to normal levels (LVL 1)
• temperature over sensor 5 (PHY) decreases back

Reverting back to 2.4.5.71 version restores normal working state with having both stack links active and FAN speed at level 1.

What I dont understand is whether FAN speed is just related to increased temperature (+15C) on PHY sensor (to 78C value) (but it shouldnt be as we are not hitting warning (95) or critircal (98) threshold), or it simply relates to fact that second 10G interface was activated... 

Unfortunately original bugreport CSCvq41525 is closed with workaround about reverting back to 2.4.x. branch, but no info about plans for fixing in future releases. 

For time being the only recommendation for similar setups is to keep your software updated to latest version of 2.4.x.x branch.