Sun SPARC Enterprise(R) M3000/M4000/M5000/M8000/M9000 Servers: Fan/fantray temperature and failure behavior

Asset ID:	1-71-1019147.1
Update Date:	2012-07-30
Keywords:

Solution Type Technical Instruction Sure

Solution 1019147.1 : Sun SPARC Enterprise(R) M3000/M4000/M5000/M8000/M9000 Servers: Fan/fantray temperature and failure behavior

Applies to:

Sun SPARC Enterprise M4000 Server
Sun SPARC Enterprise M9000-32 Server - Version: Not Applicable and later [Release: N/A and later]
Sun SPARC Enterprise M9000-64 Server - Version: Not Applicable and later [Release: N/A and later]
Sun SPARC Enterprise M8000 Server
Sun SPARC Enterprise M5000 Server
All Platforms
Sun SPARC Enterprise(R) Mx000 Servers: Fan/fantray temperature and failure behavior

Goal

This document describes the fan / fan tray redundancy, the behavior of the systems in case of fan / fan tray failure and the fan speed control depending on inlet temperature.

Solution

This section describes the environment and failure behavior.

The showenvironment command is used to display various temperatures and fan speeds.

XSCF> showenvironment
XSCF> showenvironment temp
XSCF> showenvironment Fan

The showaltitude command is used to display the altitude setting.

XSCF> showaltitude

Note: In the following tables, threshold temperatures are temperatures where an action is taken, the sign ">" means greater or equal and the sign "<" means smaller or equal in this context.

M3000

M3000 systems have a single inlet temperature sensor on the Operator Panel (OPNL)
M3000 systems have the exhaust temperature sensor on the Motherboard (MBU). The CPU chip has a temperature sensor.
Fans on M3000 systems have 10 speeds: level 1 .. level 10

Fan failure behavior

Failure of a FAN_A, and if the other FAN_A in the cooling group is operational, then this other fan's speed is raised to full speed (level 10) and the speed of all other fans on the platform is raised to high speed (level 9).
If the failure is due to a failure of a fan in a PSU, and if all the other fans in the PSUs of the cooling group are operational, then the other fan of this PSU has its speed raised to full speed and the speed of all other fans on the platform is raised to high speed.
If the second fan in a cooling group becomes non-operational, then XSCF shuts down the domain and powers off the platform.

M3000 Cooling Groups	CG#1	CG#2
Fans	FAN_A#0 FAN_A#1	Fan in PSU#0 Fan in PSU#1
Hardware	MBU DIMMs PCI slots	DVD HDDs PSUs

M3000	Typical fan speed
Fan speed	Standby	level1	level2	level3	Level4 *	level5	level6	level7	level8	level9	level10
RPM FAN_A#	0	3500	3550	3600	3700	3800	4100	4400	5000	5800	6600
RPM Fan PSU	3400	6000	6000	6000	6000	6200	6350	6550	6700	6900	12000

* Level 4 is the initial value after power on

M3000	Minimum fan speed below which a fan is declared failed.
Fan speed	Standby	level1	level2	level3	Level4	level5	level6	level7	level8	level9	level10
RPM FAN_A#	N/A	2130	2160	2190	2220	2290	2460	2670	3000	3500	3950
RPM Fan PSU	N/A	4800	4800	4800	4800	4960	5120	5280	5440	5600	9440

M3000	Fan speed relation to inlet temperature (°C) 500 m or below
Fan speed	Standby	level1	level2	level3	Level4	level5	level6	level7	level8	level9	level10
Fan speed	none	low	low	low	low	middle	middle	high	high	high	full
threshold temp.	Domain power off	< 19	> 20 < 21	> 22 < 23	> 24 < 25	> 26 < 27	> 28 < 29	> 30 < 31	> 32 < 33	> 34	*
Inlet overtemperature set: > 38 Inlet overtemperature reset: < 35

M3000	Fan speed relation to inlet temperature (°C) 501 m to 1000 m
Fan speed	Standby	level1	level2	level3	Level4	level5	level6	level7	level8	level9	level10
Fan speed	none	low	low	low	low	middle	middle	high	high	high	full
threshold temp.	Domain power off	< 17	> 18 < 16	> 20 < 21	> 22 < 23	> 24 < 25	> 26 < 27	> 28 < 29	> 30 < 31	> 32	*
Inlet overtemperature set: > 36 Inlet overtemperature reset: < 33

M3000	Fan speed relation to inlet temperature (°C) 1001 m to 1500 m
Fan speed	Standby	level1	level2	level3	Level4	level5	level6	level7	level8	level9	level10
Fan speed	none	low	low	low	low	middle	middle	high	high	high	full
threshold temp.	Domain power off	< 15	> 16 < 14	> 18 < 16	> 20 < 21	> 22 < 23	> 24 < 25	> 26 < 27	> 28 < 29	> 30	*
Inlet overtemperature set: > 34 Inlet overtemperature reset: < 31

M3000	Fan speed relation to inlet temperature (°C) 1501m to 3000 m
Fan speed	Standby	level1	level2	level3	Level4	level5	level6	level7	level8	level9	level10
Fan speed	none	low	low	low	low	middle	middle	high	high	high	full
threshold temp.	Domain power off	< 13	> 14 < 12	> 16 < 14	> 18 < 16	> 20 < 21	> 22 < 23	> 24 < 25	> 26 < 27	> 28	*
Inlet overtemperature set: > 32 Inlet overtemperature reset: < 29

* Full speed is never set based on temperature, it is only used in case of a fan failure in the cooling group.

M3000	Overtemperature behaviour (°C)
Overtemperature status	Overtemperature	Overtemperature warning	Overtemperature fail
condition inlet temperature	see tables above	-	-
condition CPU temperature	set: > 77 reset: < 64	set: > 82	set: > 102
condition MBU temperature	set: > 64 reset: < 48	set: > 69	set: > 79
Action	Set all fans to high speed	Shutdown domain then power off platform	emergency power off platform
ereport	chassis.env.temp.ot@ <location>	chassis.env.temp.otw@ <location>	chassis.env.temp.otf@ <location>

M4000 & M5000

M4000 & M5000 systems have a single inlet temperature sensor on the Operator Panel (OPNL) FRU.
There is an exhaust temperature sensor on each IOU.
Each CPU has a temperature sensor.

Fan failure behavior

Failure of a FAN_A, and if the other FAN_A in the cooling group is operational, then this other fan's speed is raised to full speed and the speed of all other fans on the platform is raised to high speed.
If the failure is due to a failure of a fan in a PSU, and if all the other fans in the PSUs of the cooling group are operational, then the other fan of this PSU has its speed raised to full speed and the speed of all other fans on the platform is raised to high speed.
If the second fan in a cooling group becomes non-operational, then XSCF sends a shutdown request to all domains in the system and powers off the system.
FAN_B are specific for M4000 platform.

M4000 Cooling Groups	CG#1	CG#2	CG#3
Fans	FAN_A#0 FAN_A#1	Fan in PSU#0 Fan in PSU#1	FAN_B#0 FAN_B#1
Hardware	MBU CPUM#0 CPUM#1 MEB#0..4	PSUs IOU	XSCFU HDD DVDU TAPEU

M5000 Cooling Groups	CG#1	CG#2	CG#3	CG#4
Fans	FAN_A#0 FAN_A#1	FAN_A#2 FAN_A#3	Fan in PSU#0 Fan in PSU#1	Fan in PSU#2 Fan in PSU#3
Hardware	1/2 MBU CPUM#0 CPUM#1 MEB#0..3 HDD#0..3	1/2 MBU CPUM#2 CPUM#3 MEB#4..7 TAPEU DVDU	PSU#0 PSU#1 IOU0 XSCFU	PSU#2 PSU#3 IOU1

M4000 M5000	Typical fan speed
Fan speed	Standby	Low *	middle	high	full
RPM Fan PSU	2400	3600	4200	5200	8400
RPM FAN_A#	0	3200	4200	5300	5900
RPM FAN_B#	5000	10000	10000	10000	12000

* low is the initial value after power on

M4000 M5000	Minimum fan speed below which a fan is declared failed.
Fan speed	Standby	low	middle	high	full
RPM Fan PSU	1520	2680	3160	4200	6400
RPM FAN_A#	N/A	2560	3280	4080	4880
RPM FAN_B#	3000	7520	7520	7520	9440

M4000 M5000	Fan speed relation to inlet temperature 500 m or below (with air filters installed, lower all threshold temperatures by 3 °C)
Fan speed	Standby	low	middle	high	full
threshold temp.	Domains powered off	< 23	> 25 < 28	> 30	*
Inlet overtemperature set: > 38 Inlet overtemperature reset: < 35

M4000 M5000	Fan speed relation to inlet temperature (°C) 501 m to 1000 m (with air filters installed, lower all threshold temperatures by 3 °C)
Fan speed	Standby	low	middle	high	full
threshold temp.	Domains powered off	< 21	> 23 < 26	> 28	*
Inlet overtemperature set: > 36 Inlet overtemperature reset: < 33

M4000 M5000	Fan speed relation to inlet temperature (°C) 1001 m to 1500 m (with air filters installed, lower all threshold temperatures by 3 °C)
Fan speed	Standby	low	middle	high	full
threshold temp.	Domains powered off	< 19	> 21 < 24	> 26	*
Inlet overtemperature set: > 34 Inlet overtemperature reset: < 31

M4000 M5000	Fan speed relation to inlet temperature (°C) 1501m to 3000 m (with air filters installed, lower all threshold temperatures by 3 °C)
Fan speed	Standby	low	middle	high	full
threshold temp.	Domains powered off	< 17	> 19 < 22	> 24	*
Inlet overtemperature set: > 32 Inlet overtemperature reset: < 29

* Full speed is never set based on temperature, it is only used in case of a fan failure in the cooling group.

M4000 M5000	Overtemperature behaviour (°C)
Overtemperature status	Overtemperature	Overtemperature warning	Overtemperature fail
condition inlet temperature	see tables above	-	-
condition CPU temperature	set: > 79 reset: < 71	set: > 82	set: > 104
condition IOU temperature	set: > 60 reset: < 49	set: > 65	set: > 75
Action	Set all fans to high speed	Shutdown all domains then power off platform	Emergency power off platform
ereport	chassis.env.temp.ot@ <location>	chassis.env.temp.otw@ <location>	chassis.env.temp.otf@ <location>

M8000 & M9000

M8000 & M9000 systems have a single inlet temperature sensor on the Sensor (SNSU) FRU.
M8000 systems have exhaust temperature sensors for each CMU.
M9000 systems have exhaust temperature sensors for each CMU and each XBU.
Each CPU has a temperature sensor.

M8000 Cooling Groups	CG#1	CG#2
Fan Trays	FAN_A#0 FAN_A#1 FAN_A#2 FAN_A#3 FAN_B#0 FAN_B#1	FAN_B#2 FAN_B#3 FAN_B#4 FAN_B#5 FAN_B#6 FAN_B#7
Hardware	CMU#0 CMU#1 CMU#2 CMU#3 DDC_A#0 DDC_A#1 XSCFU	IOU#0 IOU#1 IOU#2 IOU#3

M9000 Cooling Groups	CG#1	CG#2	CG#3
Fan Trays base cabinet	FAN_A#0 FAN_A#1 FAN_A#2 FAN_A#3	FAN_A#4 FAN_A#5 FAN_A#6 FAN_A#7 FAN_A#8 FAN_A#9	FAN_A#10 FAN_A#11 FAN_A#12 FAN_A#13 FAN_A#14 FAN_A#15
Hardware base cabinet	XBUs CLCKUs XSCFUs IOU#0 IOU#2 IOU#4 IOU#6	CMU#0 CMU#1 CMU#2 CMU#3 IOU#1 IOU#3	CMU#4 CMU#5 CMU#6 CMU#7 IOU#5 IOU#7
Fan Trays expansion cabinet	FAN_A#20 FAN_A#21 FAN_A#22 FAN_A#23	FAN_A#24 FAN_A#25 FAN_A#26 FAN_A#27 FAN_A#28 FAN_A#29	FAN_A#30 FAN_A#31 FAN_A#32 FAN_A#33 FAN_A#34 FAN_A#35
Hardware expansion cabinet	XBUs CLCKUs XSCFUs IOU#8 IOU#10 IOU#12 IOU#14	CMU#8 CMU#9 CMU#10 CMU#11 IOU#9 IOU#11	CMU#12 CMU#13 CMU#14 CMU#15 IOU#13 IOU#15

Fan failure behavior:

M8000 and M9000

The fan trays have 2 fans (FAN_B) or 3 fans (FAN_A), the fans in the fan trays are N+1 redundant, which means after the first fan failure the fan tray should be replaced as soon as possible to avoid any domain/platform outage. Fan trays are not redundant.
The fans in the PSUs do not have their speed controlled by XSCF. The fan speed on these fans is controlled by a microcontroller internal to the PSU.If a fan in a PSU fails, then the PSU is powered off and deconfigured.
If there are insufficient operational PSUs to power the platform, then the platform is powered down.

M8000 specific

A failure in cooling group #1 or cooling group #2 will affect the entire platform.
If second fan in a specific fantray becomes non-operational or if the fan tray itself fails it then XSCF will not permit to power up the platform. If the failure happens while the platform is powered up, all fan speed in the platform will be raised to high, domains will not be shut down, only warning messages will be issued. In this situation, the system relies on the overtemperature behavior described later in this document.

M9000 specific

A failure in cooling group #1 will affect the entire platform. A failure in cooling group #2 or cooling group#3 will affect only their specific FRUs, hence the domains being cooled by that cooling group.
If second fan in a specific fantray becomes non-operational or if the fan tray itself fails it then XSCF will not permit to power up the platform or specific FRUs. If the failure happens while the platform is powered up, all fan speed in the platform will be raised to high, domains will not be shut down, only warning messages will be issued. In this situation, the system relies on the overtemperature behavior described later in this document.

M8000 M9000	Typical fan speed
Fan speed	Normal	High
RPM FAN_A# FAN_B#	3700	5500

M8000 M9000	Minimum fan speed below which a fan is declared failed.
Fan speed	Normal	High
RPM FAN_A# FAN_B#	3100	4200

M8000 M9000	Fan speed relation to inlet temperature (°C) 1500 m or below
Fan speed	Normal	High
threshold temp.	< 24	> 27
Inlet overtemperature set: > 36 Inlet overtemperature reset: < 32

M8000 M9000	Fan speed relation to inlet temperature (°C) 1501 m to 2000 m
Fan speed	Normal	High
threshold temp.	< 22	>25
Inlet overtemperature set: > 34 Inlet overtemperature reset: < 30

M8000 M9000	Fan speed relation to inlet temperature (°C) 2001 m to 2500 m
Fan speed	Normal	High
threshold temp.	< 20	> 23
Inlet overtemperature set: > 32 Inlet overtemperature reset: < 28

M8000 M9000	Fan speed relation to inlet temperature (°C) 2501 m to 3000 m
Fan speed	Normal	High
threshold temp.	< 18	> 21
Inlet overtemperature set: > 30 Inlet overtemperature reset: < 26

M8000 M9000	Overtemperature behaviour (°C)
Overtemperature status	Overtemperature	Overtemperature warning		Overtemperature fail
condition inlet temperature	see tables above	-	-	-	-
condition CPU temperature	set: > 79 reset: < 71	set: > 82	-	set: > 104	-
condition CMU temperature	set: > 61 reset: < 49	set: > 66	-	set: > 76	-
condition XBU temperature	set: > 61 reset: < 49	-	set: > 66	-	set: > 76
Action	Set all fans to high speed	Shutdown domains in CG then power off all hardware in CG	Shutdown all domains then power off platform	emergency power off all hardware in CG	emergency power off platform
ereport	chassis.env.temp.ot@<location>	chassis.env.temp.otw@ <location>		chassis.env.temp.otf@ <location>

Cooling requirements:

	Rated Power	Cooling Requirements		Flow
	W	BTU/h	KJ/h	m3/min
M3000	470	1603	1692	1.75
M4000	2350	8018	8046	7
M5000	4590	16036	16524	14
M8000	10500	35857	37800	94
M9000-32	21300	72740	76680	102
M9000-64	42600	145479	153360	205

Required environment:

	Temperature (°C)			Relative Humidity (%)
	Non-Op.	Operating		Non-Op.	Operating
	Range	Range	Best Range	Range	Range	Best Range
M3000	0 to 50 -20 to 60 (packed)	0-500 m: 5-35 501-100 m: 5-33 1001-1500 m: 5-31 1501-3000 m: 5-29	21-23	0-93	20-80	45-50
M4000/M5000	0 to 50 -20 to 60 (packed)	0-500 m: 5-35 501-100 m: 5-33 1001-1500 m: 5-31 1501-3000 m: 5-29	21-23	0-93	20-80	45-50
M8000/M9000	0 to 50	0-1500 m: 5-32 1501-2000 m: 5-30 2001-2500 m: 5-28 2501-3000 m: 5-26	22-26	8-80	20-80	40-50

To discuss this information further with Oracle experts and industry peers, we encourage you to review, join or start a discussion in the My Oracle Support Community - M Series Servers

Internal section

This document describes the state as of XCP 1090

Additional troubleshooting information:

A. In case of:

multiple fan failures occurred in a short timeframe
all fans of same type rotating exactly at the same speed

please pay attention and consider a fan controller issue before replacing all the affected fan tray(s).

B. FF platforms (M5000 and M4000) have different fan backplanes:

M5000 has one Fan Backplane (for 172mm Fans, FAN_A type) that includes the fan controller
M4000 has two Fan Backplanes: one for 172mm Fans, one for 60mm Fans (FAN_B type) that includes the fan controller

Keep this in mind when dealing with multiple or repeated fan failures (i.e.: repeated FAN_A fault on M5000 may be due to failed controller, that is not included on the FAN_A backplane); check below the showhardconf output bot both platforms:

M5000

        FANBP_C Status:Normal; Ver:0501h; Serial:NN110224CD;
        + FRU-Part-Number:CF00541-3099 01   /541-3099-01          ;
        FAN_A#0 Status:Normal;
        FAN_A#1 Status:Normal;
        FAN_A#2 Status:Normal;
        FAN_A#3 Status:Normal;

M4000

    FAN_A#0 Status:Normal;
    FAN_A#1 Status:Normal;
    FANBP_B Status:Normal; Ver:0201h; Serial:BF0844MV6G  ;
        + FRU-Part-Number:CF00541-0909 04   /541-0909-04          ;
        FAN_B#0 Status:Normal;
        FAN_B#1 Status:Normal;

FAN CR's:

6875469 - On OPL M3000 Ikkaku, fan speed is set to level 4 upon domain power up, regardless of conditions.
6870490 - On M4000/M5000/M8000/M9000, fan alarm condition chgs while XSCF is down are ignored on XSCF boot

References:

OPL FF & DC Environment preso (Andre Beusch): Environment TOI
OPL DC Hardware FAQ (Dan Nygren)
Troubleshooting a Noisy FAN on OPL Servers: Doc 1339901.1
Tracking page for cases where customers complain about fans being too noisy in M4000 / M5000 systems. Noisy Fan Tracking Page.

Keywords:
OPL, Mx000, thermal temp, fan, fantray, tray, redundancy, M4000, M5000, M8000, M9000, M9000+ speed

Attachments

This solution has no attachment