On Mon, Oct 1, 2012 at 7:49 PM, Edward Austin <[email protected]> wrote:
> Hi there
>
> I have a V210 that I use at home for SW development. This is a fine
> Solaris machine and fan control is handled nicely by picl with the fans
> spooling down to a nice hum once this service starts.
>
> Wish i could say the same about a newly acquired T1000 - this machine -
> as small as a V100 - and 10X as noisy beats every server I have ever
> experienced in excessive noise - with the fans spooling up to 9K RPM and
> staying there even when picl starts from Solaris 10.
>
[SNIP]
Hi!
Yep. Everybody knows that the T1000 is a "bit" louder.
But your thread inspired me to find out, why.
It is true, that the fans run at max speed, which is 4 times as fast as
they needed to (under normal altitude /temp/load conditions).
In ALOM sc showenvironment confirmes what Edward said:
Fans (Speeds Revolution Per Minute):
----------------------------------------------------------
Sensor Status Speed Warn Low
----------------------------------------------------------
FT0/F0 OK 8776 2240 1920
FT0/F1 OK 8967 2240 1920
FT0/F2 OK 8776 2240 1920
FT0/F3 OK 8967 2240 1920
While the Hypervisor Firmware (I'm not talking about OBP [I'm not sure at
this point, if OBP does, but I moved around in the device tree and didn't
find anything suspicious]) does have knowledge about the fans, as shown
above, a booted Illumos kernel does not. And I cannot say w/o further
digging, if that's the designed behavior, or a limited one (due to
OpenSolaris' policies, where some parts simply could not be opened up due
to pending patents and other legal stuff).
I just tested our LiveDVD.
First the bad news: Noise level never ever changes, neither after POST, nor
during boot.
Here is the relevant excerpt of prtdiag -v:
============================ Environmental Status
============================
Fan sensors:
----------------------------------------------------------------
Location Sensor Status
----------------------------------------------------------------
NNL082001W:CH/FT0/F0 RS unknown
NNL082001W:CH/FT0/F1 RS unknown
NNL082001W:CH/FT0/F2 RS unknown
NNL082001W:CH/FT0/F3 RS unknown
Temperature sensors:
----------------------------------------------------------------
Location Sensor Status
----------------------------------------------------------------
NNL082001W:CH/MB/IOB T_CORE unknown
NNL082001W:CH/MB/CMP0 T_TCORE unknown
NNL082001W:CH/MB/CMP0 T_BCORE unknown
NNL082001W:CH/MB T_AMB unknown
However, unfortunately this happens, although picld is running.
And I don't see anything in SMF, that seems to control the fan speed.
Also do not know, if there may be relevant undocumented start parameters
for picld, but it would doubt it.
Here the described rest of the output:
root@MartUX_OpenIndiana_Edition:~# svcs -v|grep picl
online - 19:27:57 128 svc:/system/picl:default
root@MartUX_OpenIndiana_Edition:~# ps -ef|grep picl
root 1880 1486 0 19:30:21 console 0:00 grep picl
root 1846 1 0 19:27:57 ? 0:01 /usr/lib/picl/picld
root@MartUX_OpenIndiana_Edition:~# svccfg -s svc:/system/picl:default
svc:/system/picl:default> listprop
general framework
general/enabled boolean true
restarter framework NONPERSISTENT
restarter/logfile astring
/var/svc/log/system-picl:default.log
restarter/contract count 128
restarter/start_pid count 1845
restarter/start_method_timestamp time 1349231277.006732000
restarter/start_method_waitstatus integer 0
restarter/auxiliary_state astring dependencies_satisfied
restarter/next_state astring none
restarter/state astring online
restarter/state_timestamp time 1349231277.030179000
restarter_actions framework NONPERSISTENT
restarter_actions/restart integer
restarter_actions/auxiliary_tty boolean false
restarter_actions/auxiliary_fmri astring svc:/network/npiv_config:default
restarter_actions/maint_off integer
svc:/system/picl:default>
root@MartUX_OpenIndiana_Edition:~# uname -a
SunOS MartUX_OpenIndiana_Edition 5.11
oi_151a_Initial_SPARC_OpenIndiana__with_cheers_from__Martin_Bochnig sun4v
sparc SUNW,SPARC-Enterprise-T1000 Solaris
root@MartUX_OpenIndiana_Edition:~#
One would now think, we could set fan speed in Alom/sc, but I couldn't find
anything of interest.
Some folks on discussion boards suggested each other to replace the fans.
At first it sounds nice, but then: Why replacing them, if all we want is to
have the existing ones run at 25%!
On a side note, as always: A while ago I had the tempting idea to look if
the T1000 system board can somehow fit into an old U20 or U40 chassis. The
ps is already ATX (rather than a proprietary one with PDB as in most other
similar servers).
A XVR-300 would be added and voila: Nice sun4v workstation. One can
certainly find a silent cooler for the CPU if there is enough space in the
chassis. And the system board and hdds would get enough room and cooling
from the U20's/U40's normal system fan.
Well, nice idea: Almost! Smart as Sun sometimes was, they did not put a 1 $
USB controller onboard of that 10K$ system!, not even USB1.1 as on the
T2000.
And it would make little sense to configure serial console for kbd on a
second machine, just to use this modified T1000 as pseudo workstation (with
remote kbd).
So, until the T2000's get really cheap, this idea must wait ...
Back to the original problem.
I quickly looked, if there is some ERIE knowledge in the picld code.
Not much:
bash-4.1$ grep -n T100 ./usr/src/cmd/picl/plugins/sun4v//*
bash-4.1$ grep -n T100 ./usr/src/cmd/picl/plugins/sun4v//*/*
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.c:31: * Erie platforms
(T1000) do not have ok-to-remove LEDs
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.h:54:#define
ERIE_PLATFORM "SUNW,Sun-Fire-T1000"
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.h:55:#define
ERIE_PLATFORM2 "SUNW,SPARC-Enterprise-T1000"
bash-4.1$ grep -n T100 ./usr/src/cmd/picl/plugins/sun4v//*/*/*
bash-4.1$ grep -n T100 ./usr/src/cmd/picl/plugins/sun4v//*/*/*/*
grep: can't open ./usr/src/cmd/picl/plugins/sun4v//*/*/*/*
bash-4.1$ grep -n ERIE ./usr/src/cmd/picl/plugins/sun4v//*/*/*/*
grep: can't open ./usr/src/cmd/picl/plugins/sun4v//*/*/*/*
bash-4.1$ grep -n ERIE ./usr/src/cmd/picl/plugins/sun4v//*/*/*
bash-4.1$ grep -n ERIE ./usr/src/cmd/picl/plugins/sun4v//*/*
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.c:386:
((strcmp(platbuf, ERIE_PLATFORM) == 0) ||
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.c:387:
(strcmp(platbuf, ERIE_PLATFORM2) == 0)))
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.h:54:#define
ERIE_PLATFORM "SUNW,Sun-Fire-T1000"
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.h:55:#define
ERIE_PLATFORM2 "SUNW,SPARC-Enterprise-T1000"
bash-4.1$ grep -n ERIE ./usr/src/cmd/picl/plugins/sun4v//*
bash-4.1$ grep -n ERIE ./usr/src/cmd/picl/plugins/sun4v//*gedit
./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.c:31:^C
bash-4.1$ vi ./usr/src/cmd/picl/plugins/sun4v//piclsbl/piclsbl.c
[SNIP]
static void
piclsbl_init(void)
{
char platbuf[SYS_NMLN];
/* check for Erie platform name */
if ((sysinfo(SI_PLATFORM, platbuf, SYS_NMLN) != -1) &&
((strcmp(platbuf, ERIE_PLATFORM) == 0) ||
(strcmp(platbuf, ERIE_PLATFORM2) == 0)))
return;
/* retrieve the root node for lookups in the event handler */
if ((ptree_get_root(&root_node)) != NULL)
return;
/* load libpcp */
if (load_pcp_libs()) {
syslog(LOG_ERR, "piclsbl: failed to load libpcp");
syslog(LOG_ERR, "piclsbl: aborting");
return;
}
/*
* register piclsbl_handler for both "sysevent-device-added" and
* and for "sysevent-device-removed" PICL events
*/
(void) ptree_register_handler(PICLEVENT_SYSEVENT_DEVICE_ADDED,
piclsbl_handler, NULL);
(void) ptree_register_handler(PICLEVENT_SYSEVENT_DEVICE_REMOVED,
piclsbl_handler, NULL);
}
[...]
So, the strange thing: google indicates, that _some_ rare users did have
silent T1000's.
It can only be explained, if they somehow were running either extra FW,
extra OBP or extra patched picld related modules. I have no clue. I own my
T1000 only since July. And I paid just 149 EUR for it :)
-
%martin
http://wiki.openindiana.org/oi/MartUX_OpenIndiana+oi_151a+SPARC+LiveDVD
http://www.youtube.com/user/MartUXopensolaris
http://www.facebook.com/pages/MartUX_SPARC-OpenIndiana/357912020962940
https://twitter.com/MartinBochnig
http://www.martux.org (new page not yet online, but pretty soon)
Forwarding David Halko's: http://svr4.blogspot.com/
-------------------------------------------
illumos-discuss
Archives: https://www.listbox.com/member/archive/182180/=now
RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be
Modify Your Subscription:
https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4
Powered by Listbox: http://www.listbox.com
