Sun Fire v440 Over heat Problem.

UNIX Solaris
1

Dear Team,

I need some expert advice to my problem.

We have a Sun Fire v440 in our customer Place. Server is working fine and no hardware deviations are found except one problem that processors generating too much heat. I have verified and found that the room temperature was 26-27 degree.

The tempearatures of the processors are also flctuating over the day. If we put a Fan behind the server,the temp decreases to 2-3 degrees . I think due to excess room temp,this overheat issue arises. Is my assumption correct or is there any other factorsare there which related to the overheat issue. Please suggest.

NOTE: No errors found in messages files also

i am sending the logs of the server for better understanding. these are the readings taken after 2 hours duration.

chk_server.sh Wed Dec 28 08:00:00 IST 2011

Wed Dec 21 08:00:00 IST 2011

############# Server Temperature ##############

c0_p0_t_cor | 94
c1_p0_t_cor | 89
c2_p0_t_cor | 93
c3_p0_t_cor | 96
c0_t_amb | 31
c1_t_amb | 31
c2_t_amb | 31
c3_t_amb | 31
scsibp_t_am | 29
mb_t_amb | 35

chk_server.sh Wed Dec 28 10:00:00 IST 2011

############# Server Temperature ##############

c0_p0_t_cor | 91
c1_p0_t_cor | 87
c2_p0_t_cor | 90
c3_p0_t_cor | 93
c0_t_amb | 28
c1_t_amb | 28
c2_t_amb | 28
c3_t_amb | 29
scsibp_t_am | 27
mb_t_amb | 33

Thanks
Sudhansu

2

From the speficiations for sun fire v440:

And yes, the warm room is elevating the operating temperature. If the room itself experiences more temperature increases, you will have issues with overheating in the future.

I would work on the environment. Move the v440 into a rack or mount with forced circulating air. Try clearances all around for ventilation - or forced air. Maybe your fan trick will work long term - I do not know. But if your fan dies the v440 may die soon after.

We have 4 servers in a room that never exceeds 20C, and we force cool air into the rack mounts. It just isn't worth the problems caused to application data to experience a completely preventable heat outage. My opinion.

3

Thanks Jim for your views on this.

4

Dear Sudhansu,

The temp that you have mentioned is high . It should be around 19 to 20 degrees.

So Try to put coolers facing your server , that will work good.

Rgds rj

5

Thanx jegaraman,

I am communicating the same thing to customer to bring the temp of the room below 22 degree. But they suspect to change the heat sink of the processor ? Is this a solution ?? I don't think so.

In that DC they are having some SF 280R servers also. But their concern is why the temp of the SF v440 server is increasing why not with 280 servers.

Is the amount of heat generated by a SF v440 server is same as SF280R server ?????

regards
Sudhansu

6

i have never heard of a heat sink degrading but apparently it does happen (see article on heat sink degradation) ... if the issue only started recently without any changes to the environment -- which judging from the servers you mentioned (v440 and v280 both of which are already out of life) that may have already been there for years -- it is very possible that heat sink degradation could be the cause of the overheating ...

hopefully your customer has sun/oracle support or has access to a cpu parts supplier ...