Koozali.org: home of the SME Server
Obsolete Releases => SME Server 7.x => Topic started by: icpix on September 19, 2006, 06:45:43 PM
-
I think both Xeons were being used before but now I'm really
not so sure that the second is now in use. I used to get graphs
for both CPUs' activity using the sysmon contrib but now I don't.
When I put a second CPU-intensive task on I see that the first
task loses percentiles. The physical heat given off seems amiss.
Something has changed but I can't put my finger on it;~/
'htop' shows only one discrete set of figures for CPU.
----best wishes, Robert
postedit: to include the word SOLVED;~}
-
What does cat /proc/cpuinfo show?
edit-
And you could use the sysstat package which will give you a command called mpstat which will show if 2 cpu's are in use.
-
[root@teri mnt]# cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 4
model name : Intel(R) Xeon(TM) CPU 3.40GHz
stepping : 3
cpu MHz : 3402.372
cache size : 2048 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm pni monitor ds_cpl est cid xtpr
bogomips : 6806.37
[root@teri mnt]#
-
[code]
[root@teri mnt]# cat /proc/cpuinfo
processor : 0
Now I've never had a 2 cpu machine myself but shouldn't the result of that command also show processor :1 ?? What kernal you using to boot up into, SMP??
-
byte----
Give me a break;~) Who's asking the questions?
AFAIK it just tells me I've got precisely '0' CPUs running...
Something is amiss.
----best wishes, Robert
-
byte----
Give me a break;~) Who's asking the questions?
Well without more questions asked by me or someone else how do you expect to get an answer?
AFAIK it just tells me I've got precisely '0' CPUs running...
The "0" shows you that's basically the first processor installed if you had "0" then "1" then that would say 2 processors are installed.
If you don't wish for me to ask more questions then I shall stop responding to this thread
-
byte----
Didn't see the rest of the question at first.
SMP ( used to be and I should still be ) but where do I find out?
I've tried STARING at the darned monitor at boot but miss it every time.
Is that fact in the logs, I've looked through dmesg.
(posts are crossing and being edited)
[root@teri mnt]# mpstat
Linux 2.6.9-34.0.2.EL (teri) 09/19/2006
06:03:15 PM CPU %user %nice %system %iowait %irq %soft %idle intr/s
06:03:15 PM all 20.29 0.00 1.52 0.19 0.04 0.00 77.96 1052.41
[root@teri mnt]#
Yes, of course, I do wish you continue;~0 I was being ironic earlier;~)
----best wishes, Robert
-
[root@teri mnt]# mpstat
Linux 2.6.9-34.0.2.EL (teri) 09/19/2006
That seems to say your using single processor as if you type...
[root@fun ~]# uname -a
Linux fun 2.6.9-34.ELsmp #1 SMP Wed Mar 8 00:27:03 CST 2006 i686 athlon i386 GNU/Linux
[root@fun ~]#
Notice the SMP on the end of mine.
Yes, of course, I do wish you continue;~0 I was being ironic earlier;~)
OK we are relpying and posting pretty quick :shock:
-
byte----
[root@teri mnt]# uname -a
Linux teri 2.6.9-34.0.2.EL #1 Fri Jul 7 19:24:57 CDT 2006 i686 i686 i386 GNU/Linux
[root@teri mnt]#
Something has gone very wrong. I am now certain, thanks to your help,
that I am down to one processor. SME7 just isn't recognising the 2nd CPU
even at boot, so presumably it won't at any other time either. I mean it's
got to be using both CPUs during the boot to be able to load the appropriate
'kernel'? Something has gone awry a treat;~|
----best wishes, Robert
-
byte----
If one processor gets an unfair load it overheats...
upper CPU was 50C whereas the lower CPU was 38C
...temperatures from BIOS (as I found out with that reboot).
A somewhat long story radically pruned...
It was a tedious hardware build.
SME7 beta's kernel panic'd until I disabled ACPI then was OK.
Relatively SURE at that point both CPUs ran and SME7 was ELsmp.
Certain I got sysmon graphs for CPU0 and CPU1.
Found out the hard way that Asus motherboard's LMxx stuff wasn't in
the Linux kernel and so CPUeither/both temperature monitoring was u/s.
SME7 gold came and went.
YUMs came and went.
During an area-wide power cut found out that the absence of ACPI
meant automatic shutoff did not occur but thought little (trusting to RAID).
By and by I tried switching the BIOS ACPI back on... boot 'worked'.
I can't honestly remember seeing or even checking for EL or ELsmp...
Nowadays I see no graphs at all for ANY CPU in sysmon.
htop shows only one CPU.
Then you showed me some interrogative commands.
Now I KNOW my SME7 is using only the upper processor.
From the heat signature this is the one the box uses when in the BIOS menu.
Is there some crowbar way of forcing SME7 to adopt both processors?
So, crux question, did my attempt to turn ACPI back on disturb something,
on a one-way basis, in the permanent 'setup' of how SME7 boots/works?
My feeling is that it did and that I am likely to have to re-install - probably
from new.
Your thoughts would be greatly appreciated;~)
----best wishes, Robert
-
Robert,
I would think that what may have happened is that if you say you've been doing YUM updates which could include kernel updates is when it reboot GRUB automatically choose what it thought was the best kernel to use and that's why you only see one CPU, have you tried booting into ELsmp mode? What should happen is when you select the ELsmp boot from the bootloader menu it will automatically use that kernel from now on unless it/you changes it.
-
icpix
Its just like what byte is saying ...
Something has gone very wrong. I am now certain, thanks to your help,
that I am down to one processor. SME7 just isn't recognising the 2nd CPU
From your snipit [root@teri mnt]# uname -a
Linux teri 2.6.9-34.0.2.EL #1 Fri Jul 7 19:24:57 CDT 2006 i686 i686 i386 GNU/Linux
[root@teri mnt]#
It says clearly that you have booted up in single CPU mode.
On bootup just select smp and see what the output is.
Regards,
Tib
-
byte/tib----
Yes!!!
I had a very strange YUM episode a while ago.
A massive load of changes suddenly became 'apparently' necessary.
So many that I thought to go to the mirrors to pick up a new ISO...
but there was no new ISO... when I attempted to do the YUM something
happened (I don't know what) and suddenly there were no further YUM
apparent requirements. Since then YUM has brought no more changes.
I think it is possible that that YUM episode and the necessary subsequent
reboot afterwards 'reset' my ELsmp to EL and from thereon it was fixed.
Exactly how do I force a 'ELsmp boot from the bootloader'?
Is that a reboot and a function key interrupt during POST or something?
What precipitates that sort of interaction?
I *MUST* be *VERY* sure it will not force a rebuild that wipes the drives.
I can or will nuke SME7 (easy to rebuild) but *NOT* the mounted h/w
RAID5 array (2TB).
----best wishes, Robert
-
byte/tib----
Exactly how do I force a 'ELsmp boot from the bootloader'?
When the you reboot you will see a white menu with a penguin on it that it's what we refer to as a bootloader, in that bootloader you should see the ELsmp option and once select it will remeber that option next time you go for a reboot.
I *MUST* be *VERY* sure it will not force a rebuild that wipes the drives.
I can or will nuke SME7 (easy to rebuild) but *NOT* the mounted h/w
RAID5 array (2TB).
I'm not sure what it would do there, so if you was to try it - its down to you as I don't use RAID 5 and never have.
-
I found switching between EL and SMP it won't do anything with the raid ...
My test server is setup in software raid5 and have tried that already ... all data was ok and no rebuild or anything took place.
I had the same problem on update that it went to EL and not SMP ... all should be cool once you re-boot and select SMP on the boot screen ... just watch as it boots cause the boot loader doesn't stay up for long.
Regards,
Tib
-
byte/tib----
Production box is a separate SME6 (ie this comms channel!).
I have SME7 on hold in the boot showing the white penguin.
There are FOUR options: two for CentOS and two for SME Server...
CentOS 2.6.9-34.0.2 EL
CentOS 2.6.9-34.0.2 ELsmp
SME Server 2.6.9-34.EL
SME Server 2.6.9-34.ELsmp
...and I think I am currently 'running' in the first option.
Whereas I should be the last option? SME Server smp?
[postedit: I typed up the order slightly wrong... see later posting]
Think I know (more) what happened when things went wrong.
It could have been as we have discussed... but also the site's
one monitor and keyboard could have been on the SME6
which has old-age problems (I built it 10yrs ago!) and the
SME7 YUM reboot might have gone through blind (no monitor)
tap tap etc...
hence the 1st option?
hence CentOS non-smp?
...maybe.
Incoming utility power is dipping already.
A storm is coming... it's now after 11pm - could be a long night;~)
----best wishes, Robert
-
.
There are FOUR options: two for CentOS and two for SME Server...
CentOS 2.6.9-34.0.2 EL
CentOS 2.6.9-34.0.2 ELsmp
SME Server 2.6.9-34.EL
SME Server 2.6.9-34.ELsmp
...and I think I am currently 'running' in the first option.
Whereas I should be the last option? SME Server smp?
You are actually running the third option "SME Server 2.6.9-34.EL" and yes you will need the last option "SME Server 2.6.9-34.ELsmp"
Think I know (more) what happened when things went wrong.
It could have been as we have discussed... but also the site's
one monitor and keyboard could have been on the SME6
which has old-age problems (I built it 10yrs ago!) and the
SME7 YUM reboot might have gone through blind (no monitor)
tap tap etc...
hence the 1st option?
hence CentOS non-smp?
...maybe.
Have you reported to the bug tracker? or checked the bug tracker as I know a recent kernel update has been sent out via updates.
-
byte----
Typed up in the right order (as displayed on the monitor)
There are FOUR options: two for CentOS and two for SME Server...
CentOS 2.6.9-34.0.2 EL
CentOS 2.6.9-34.0.2 ELsmp
SME Server 2.6.9-34.ELsmp
SME Server 2.6.9-34.EL
I believe I was in the CentOS non-smp option.
Have now selected the actual third option...
SME Server 2.6.9-34.ELsmp
...notice it does not have the -0.2 that my snippets showed earlier.
htop - shows CPU1 and CPU2 figures ;~))
sysmon - shows CPU1 and CPU2 graphs ;~))
dmesg - first line shows smp ;~))
[root@teri /]# uname -a
Linux teri 2.6.9-34.ELsmp #1 SMP Wed Mar 8 00:27:03 CST 2006 i686 i686 i386 GNU/Linux
[root@teri /]#
[root@teri /]# cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 4
model name : Intel(R) Xeon(TM) CPU 3.40GHz
stepping : 3
cpu MHz : 3402.234
cache size : 2048 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm pni monitor ds_cpl est cid xtpr
bogomips : 6806.77
processor : 1
vendor_id : GenuineIntel
cpu family : 15
model : 4
model name : Intel(R) Xeon(TM) CPU 3.40GHz
stepping : 3
cpu MHz : 3402.234
cache size : 2048 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm pni monitor ds_cpl est cid xtpr
bogomips : 6799.37
[root@teri /]#
Thanks to you all
we're 'cooking on gas'...
it's a strange/English TV advert
...as opposed to cooking a CPU by overheating it.
I will amend the thread's title to include [SOLVED];~)
----best wishes, Robert
-
For anyone researching this topic
See additional thread here:
http://forums.contribs.org/index.php?topic=33825.0
and bug report here:
http://bugs.contribs.org/show_bug.cgi?id=1930
Royce Holdeman