Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie
HP Proliant ML310e Gen8 - Home use
Options
-
26-10-2013 10:31am#1I have acquired a ML310e and have never had to deal with anything like this previously, so am a little at sea, and looking for guidance.
I am reasonably familiar with PC hardware.
The box has 1 HDD and had RAID turned on.
I installed Linux to the one HDD, but could not boot ..... ok, I figured because it was in some form of RAID array set up or something.
I could boot from USB stick and CD.
I turned off the RAID functions, reinstalled Linux, and am able to boot.
Everything working great ..... didn't do a lot ..... a small bit of tweaking in the set up is all.
It ran almost silently even while compiling drivers, keeping the CPU temp around the 30 C mark at idle ..... rising to 38 C max during my use.
OK ....... that's the background!
I installed a DTT tuner card into one of the available PCI-e slots ..... and now the fans are running a lot faster, making a lot more noise and reducing the CPU temp to about 24 C at idle. The 'load' the tuner card puts on the CPU is minuscule.
If I disable the card the fans return to 'normal' near silent running.
So my questions are ......
are the fan speeds controlled by the 'sensed' load on the PSU or by temp sensors?
is this expected behaviour? If yes can it be changed?
Because of the availability of the PCI-e slots, and the obvious power of the CPU (less important) I had thought to have this box running as a file sharer but also to do media server using a couple of PCI-e tuner cards.
As mentioned, this is my first steps into dealing with a 'proper' server box, so I realise that I am almost certainly missing some simple piece of info which would help.
All suggestions, info, relevant links appreciated.
Thanks for reading
0
Comments
-
How do you disable the card? Remove external power from it?
IIRC speeds are controlled by various temperature sensors and the bios. Using the temp sensors should allow it ramp up and down according to temp and the bios just sets the minimum fan speed so you could peg it to 100%. I don't believe the fans speeds are PSU load based.
I had similar with 2 Ml110G7's a couple of years back but can't remember the fix
. I had 2 identical setups but one server ran its fans at 100% after a PCI card was installed.
What Linux distro are you running and have you updated it?
Is the card detected by the O/s?
The fan noise starts almost immediately after powering up right?
To solve it on the 110's I would of done the following
Installed the latest PSP and Firmware from HP - Does your have an Intelligent provisioning option on boot if so it can be launched from there. If not download the firmware update dvd and the PSP for Linux
Tried different PCI slots.
Try a different PCI card and see if the problem reoccurs - Doesn't matter what it is you just want to populate the slot.
Reinstall the O/s with the card present from my reading I think you installed the O/s first?
I'm starting to think the reinstall with card present was the fix. I've a vague recollection of have both ML110's but only a single card and I installed the second when it arrived and the problem occurred the only difference between the 2 being one was installed with it present and one without.0 -
How do you disable the card? Remove external power from it?
There is a setting in BIOS to disable a populated PCI-e slot.
Doing so reverts the fan behaviour to that of having no card in the slot.IIRC speeds are controlled by various temperature sensors and the bios. Using the temp sensors should allow it ramp up and down according to temp and the bios just sets the minimum fan speed so you could peg it to 100%. I don't believe the fans speeds are PSU load based.
That is what I would expect of a standard PC, but this is obviously different in some way.I had similar with 2 Ml110G7's a couple of years back but can't remember the fix . I had 2 identical setups but one server ran its fans at 100% after a PCI card was installed.
The fan behaviour during POST is as expected ..... it ramps up like a jet engine taking off and then settles back.
With the PCI-e card disabled or withdrawn, it continues to throttle back until it is barely audible.
With the card inserted and enabled, it does not continue to throttle back, and remains running at reasonable speed which is quite noisy, but is nowhere close to the max it gets to during POST.What Linux distro are you running and have you updated it?
Is the card detected by the O/s?
The fan noise starts almost immediately after powering up right?
The fan behaviour is present prior to booting any OS.To solve it on the 110's I would of done the following
Installed the latest PSP and Firmware from HP - Does your have an Intelligent provisioning option on boot if so it can be launched from there. If not download the firmware update dvd and the PSP for Linux
There is an Itelligent Provisioning option, but I know nothing about it, except I believe I turned it off.
PSP is operating system software, but the problem exists prior to any OS booting.
I am not at all sure about firmware.
This box is new, so there should be very little to update (says he hopefully).Tried different PCI slots.
Done ..... no difference that I could hear.Try a different PCI card and see if the problem reoccurs - Doesn't matter what it is you just want to populate the slot.
Tried three ...... two tuner cards, and both had the same effect.
I also tried a graphics card.Reinstall the O/s with the card present from my reading I think you installed the O/s first?
I'm starting to think the reinstall with card present was the fix. I've a vague recollection of have both ML110's but only a single card and I installed the second when it arrived and the problem occurred the only difference between the 2 being one was installed with it present and one without.
I did reinstall the OS with the card/s present but that made no difference.
In any case the 'problem' arises prior to booting any OS, but does continue when the OS is booted.
******
In order to better explain I have just powered up the box without any PCI-e card ........
During POST the fan is run at high speed for a short burst, and then settles at some (what I will call) 'mid-point' speed. This mid-point is audibly noticeable.
As POST progresses and the box enters the HP Initialisation phase, at some point during that the fan begins to slow and settles at a low speed which is quite capable of keeping the CPU cooled.
**************
With a PCI-e card inserted and enabled, the Initialisation seems to change, in that the fan does not begin to slow from its 'mid-point' speed, but stays there.
If the PCI-e card is disabled or not inserted, everything works as I would expect (bearing in mind my lack of experience with such hardware).
Thinking about it now, it seems that the insertion of a PCI-e card changes some parameters of the Initialisation which in turn causes this phenomenon ....... that is just me trying to guess at a cause
Thanks for the response
0 -
-
-
-
Advertisement
-
Firmware d/load in progress ...... failed at first attempt ...... needed eth connection to the other NIC and then it connected to the remote server.
I have a passively cooled graphics card in Slot 4 at present, and the fan is running too fast (~ 'mid-point' as described above).
Maybe a firmware upgrade will cure this ......
**
Yes it is damned slow, whatever it is downloading!
I didn't have to register, just accept the EULA.
When this is finished I will have to look at other options there to see if anythig is applicable to home use of this server box.
**
Thanks for the responses
.0 -
-
-
-
-
Advertisement
-
Last test ....... tried changing from the AHCI controller to the B120i RAID Controller ......... no difference at all .......
...... the fan behaviour changes when the PCI-e card is enabled.
With this behaviour this machine is not at all suitable for the purpose to which I wished to put it :mad:
0 -
Try installing a trial of Server 2012 using Intelligent Provisioning to install rather than booting from the media you've nothing to loose. It's a supported O/s so may well put manor on the fan speed once booted if it does at least you'll know whats happening and try compile drivers for your kernel.
Sorry, I don't quite follow the reasoning.
The fan behaves 'properly' without a card enabled and 'improperly' with the card enabled, in the absence of any OS.
You seem to suggest that the fan speed should be controlled by the OS ...... yet the choice in the BIOS is for the HP firmware to control power management etc. etc.
Do you know of anything from HP which addresses - or comes near to addressing - this issue?
...... maybe I have a faulty unit, if no one else has reported/commented upon this phenomenon .....
*****
EDIT:
I will try using Int. Prov. to install an OS and see what happens. It appears I will first have to download a suitable ISO.
The OSs available doing things this way are very limited ..... I looked at Red Hat Linux, but it seems I would have to have an account or something to download it.
I will try a basic version of CentOS first I guess, and see what happens.
If I can overcome this fan problem, this machine looks like it will do everything I want and a lot more.
Technically it is far beyond me ..... maybe if I had a few years exposure to servers things would be different
.0 -
-
My only reasoning is that I had a couple of identical ML110 G7's and one experienced a similar fan noise issue after a PCI card was installed. Both where side by side so it was very obvious that despite the same hardware and config one was noticeably louder. The only difference was that I installed the PCI card in the second after the O/s was installed as it wasn't available when I was building the machine.It's 18 months or so ago so I can't be sure but I believe the fix was to install the O/s with the card in place.
OK, understand now.
I am in a bit of a pickle ...... the tuner cards are in use presently in another machine, so all I get is some test time at times when they are not requiredYou can get an ISO for 2012 by registering here http://technet.microsoft.com/en-US/evalcenter/dn205286.aspx At the very least you'll have a supported O/s installed and you can call HP to get their advice. You never know they might ship a new mainboard or class the unit DOA.
Thanks I must try that ...... but it will require quite a bit of arranging ....... and the fact that I have not used any Win product since W2K won't be helpful either
This Gen8 has something they call 3D Sensor Array or something like that ...... which seems to take power and temp readings and then decide on cooling parameters (presumably including fan speeds).
I don't know how to get access to that or its settings.
Using this machine has a rather large learning curve for me --- due to specific firmware with which I am completely unfamiliar I suppose.
If I could disable all that, treat it like a standard desktop, and then enable some parts as required, it would make it easier on me
I am impressed with the hardware though and how it is assembled.
****
On another note, I have found a lot of info pages about various aspects of HP servers and the Gen 8 in particular too, but most, if not all of it, presumes familiarity with things servers and HP servers in particular.
I would like to find something very basic relating to the Gen 8 servers ...... things like how to do set ups and so on, and what might be the expected results from making changes to some functions etc .... but written for the likes of me who have no prior knowledge ...... every time I meet an acronym I have to search for its meaning in relation to Gen 8.
I really do need to find a simple overview to 'get a handle' on this.
Thanks for the help
.0 -
D/loading Win Server 2012 R2 evaluation as I write.
I had tried CentOS but it was an unsupported OS so it just bunged me back to the CD for installation.
Maybe the behaviour with Win will be different.
Another learning experience for me ..... installing Win
That will have to wait until the tuner cards can be made available ...... or maybe I will use one of them only for this testing.
That should be sufficient I imagine.
...... the saga continues ........
.0 -
-
-
Make up a key (25 digits) the Intelligent Provisioning doesn't validate it and the trial shouldn't require one
Yes, I did just put in a made up one and it installed (or seemed to) but could not boot ....... error in BOOT\BSD or something similar ..... and the DVD could not repair it.
I found no means of getting a valid install through Intell. Prov.
I was able to install manually, but the fans were noisy as before .... or maybe even worse.
I tried finding my way around the OS, but did not have great luck
The tuner cards require drivers and there are several available ...... XP/Vista, Win7 and even Win8 I think.
I am unsure which of those would be most suitable ..... the interface looks a lot like pics I have seen of Win 8 ....
In any case, no real progress on the fan problem.
I will maybe have to look to registering the product and ringing HP during the week to see if there is something I am missing that could be done about the fan speed & noise.
Hopefully there will be something available.
I do recall that the sensors detection I used on Linux did not see any fan speeds reported, but did report the CPU temps.
Thanks again for the suggestions, but honestly I am flying completely blind with Win and have no idea what might be possible there.
regards.0 -
Got into the iLO settings from another PC.
The CPU temp is reported at 40C and never changes that I can see.
Without the PCI-e cards installed the fans run at
Fan1 8%
Fan2 9%
With the cards installed both fans run at
26%
I found no way to alter the 'trigger' settings of the fan speeds.
I also have no idea why the CPU should be reporting the same temp regardless the fan speed ..... both tests done with machine sitting idling for 10 mins at the bootloader (Grub) prompt.
Checking the CPU temps in PCLOS running Live shows a marked difference between having fans at 26% and fans at 8%.
It looks like a call to the HP help desk is in order 0 -
-
Advertisement
-
That's a PITA 2012 not booting after install from Intelligent Provisioning you could try change the controller mode to the opposite to what it is now and try again.
The CPU temp is stable as the fan you're hearing is the chassis not CPU fan which I'm guessing it pretty constant.
No, the fan is exhausting the air through the CPU heatsink ..... there are two fans only and both go to 26% according to the report.
In any case the fan which is exhausting from the CPU is most definitely ramping up and blowing hard out the rear of the chassis.
I have wiped the Win install ..... way too difficult for me to manage, and as I will never use it long term there is no real point in trying to learn all its 'ins & outs', and spending the few hours installing again.
Sorry, but that is somewhere I don't intend to go again ..... willingly
These pics should show you how the air-flow is managed
0 -
-
Post back how you get on with HP support I'm guessing they'll blame the card as being unsupported. What are you using the box for? A media center?
That was the intention ....... but unless this is resolved I am not sure.
So you reckon only a certain number of PCI-e cards are 'supported' in this machine?
If they come up with that story I will have a wee bit to say ..... one I am not looking for support for anything they did not build ...... and two, where is the list of 'supported' cards?
Putting a graphics card into the PCI-e graphics card slot (x 16) causes the same behaviour.
Neither is it OS dependent, as it happens without any OS present.
Then they have to explain how their firmware sees the CPU at 40C regardless its actual temp or fan speed ..... and it never moves apparently.
I won't be in much humour for excuses when I get through -- if I do --- and blaming everyone else instead of admitting there is a problem will not be acceptable.
...... and the saga moves on some more ........
.0 -
An update ..... I spent a full 3 hours on webchat with a HP rep.
After the usual attempts to blame something else and to say it is design and so on, he did begin to delve into the problem a little.
I concentrated on the iLO reporting of the two temps - CPU and chipset at 40C & 44C respectively - regardless of conditions such as load, fan speed or anything else.
Eventually he reported after a 15 min absence that all such servers they checked had the same readings .... this was after I asked him to give me instructions how to cause those temps to change. He never did try that one.
So, it seems, that iLO mis-informs the admin of the server. Two of the reported values are not actual values at all, but are fixed and unchanging.
One of those two is the CPU temperature!!!!!
That being so, the CPU temp cannot be managed based on its temperature. I have no idea what else could be used that would make any sense!
Also ..... and this might be a greater worry ..... it also appears that all of the servers of this line they checked had the same two values.
That to me means it is not a problem with my particular server, but is a problem right across the whole line!
Makes me wonder how many admins made decisions based on fraudulent information presented to them in the iLO.
Yes it is fraudulent and not just faulty.
They have put in values representing them as actual values, knowing those values to be false (most of the time .... I suppose sometime the CPU & chipset will actually agree with the fixed values).
Because I was able to read the CPU temps when running PCLinuxOS from LiveUSB through lm_sensors, I am satisfied that the sensors are present and working for the CPU.
Unfortunately it seems HP is ignoring such readings.
Complicating matters is the fact that both main fans ..... for the CPU zone and the expansion card zone ..... seem to be tied together regarding speed.
If one ramps up, so does the other.
Which seems to mean that if I insert a PCI-e card, that fan is driven a little harder because of a need to cool the new card .... with the effect that the CPU fan also ramps up to unnecessarily cool the CPU.
I have had CPU temps at 24C in such circumstances. That is ridiculous!
It seems there is no solution for this mess except a firmware update to fix the temp monitoring and the separation of the fan controls.
I did get a promise from the HP rep (a pleasant person BTW) that the issue would be elevated to the next level, and I should hear from another rep on the matter within the next few days.
It is my fear this issue will be ignored or even worse claimed as a deliberate design.
Anyway ....... so the saga pauses, if not ends. A great hardware design let down by faulty firmware
.0 -
I have continued to try to find some factual accurate information on line about HOW the fan speeds are managed with the varying temperatures of each 'zone'.
So far I have found very little indeed.
The only references I have found have been in the various HP blurbs about the Sea of Sensors being used to manage the temps in each zone by varying the fan speeds, thus preventing unnecessary cooling of the CPU if the temp in another zone rises.
Clearly this is not what is happening here.
Simply plugging in a PCI-e graphics card (in the PCI-e zone of course) causes the CPU fan speed to rise, thus over cooling the CPU and wasting energy ........ and this happens without any appreciable rise in temperature in the PCI-e zone.
Now I *could* accept that to have things work properly a 'supported' PCI-e would have to be used, and they just *might* ramp up the speed of the PCI-e zone fan if an unsupported PCI-e card is installed, as a precautionary measure for that zone.
That *should* not cause the fan in the CPU zone to also ramp up ...... but it does!
I did come across this pic which was shown when discussing a different problem about other sensors not showing temperature thresholds (the two marked in the pic).
What I found *most* interesting is the entries for the two temps I have been concerned about. In this pic those two are also shown as 40C & 44C; the same 'apparent' readings that are fixed in my machine.
As the subject server in the pic is a different series to mine, it would appear that HP are not reporting the actual readings of those two temps in more than one series of servers!
That is really weird!
Have a look at the readings of sensors No 2 & 13 in this pic!
It is a CL380 server!
http://h20565.www2.hp.com/portal/site/hpsc/template.BINARYPORTLET/public/kb/docDisplay/resource.process/?javax.portlet.sync=530e44a3162911b1711a4d6771911c01&javax.portlet.tpst=ba847bafb2a2d782fcbb0710b053ce01_ws_BI&javax.portlet.prp_ba847bafb2a2d782fcbb0710b053ce01=wsrp-navigationalState%3DdocId%253Demr_na-c03756442-1%257CdocLocale%253D%257CcalledBy%253D&javax.portlet.rst_ba847bafb2a2d782fcbb0710b053ce01=wsrp-url%3Dhttp%253A%252F%252F16.197.209.193%253A50002%252Fsp4tssearch%252FimageServlet%253FDOCID%253Demr_na-c03756442-1%252Fc03756525.jpg%26wsrp-navigationalState%3DdocId%253Demr_na-c03756442-1%257CdocLocale%253D%257CcalledBy%253D%26wsrp-requiresRewrite%3Dfalse&javax.portlet.begCacheTok=com.vignette.cachetoken&javax.portlet.endCacheTok=com.vignette.cachetoken
I have received feedback from a user of the CL380 boxes who said he always had a problem with fan noise and this issue might help explain it ......
..... surely someone, somewhere has raised problems with this control system before now ..... I wish I could find something to indicate they had ....... either I am totally and completely wrong and misunderstanding what is going on (and if so then I *should* be able to find an explanation that would correct me), or else HP have seriously screwed up the firmware control across a lot of servers .....
Informational links and references that relate to this would be appreciated.
Thanks.
*******
EDIT for added info .....
Yesterday I ran the server with an installed OS and no PCI-e cards.
The fans operated at 8% regardless what was being done in the OS.
When idling the CPU temp was in the mid twenties centigrade.
It rose about 10 to 12 degrees C when playing a video ...... using the on chip graphics capability.
No change in fan speeds.
So I did some heavy lifting with it to observe the changes.
I did a few remasters of the installed OS, which involved mkisofs and related functions.
Both cores reached 50C with one going a degree or two higher at times.
No reaction from the fan/s.
No change at all.
The two temp readings in iLO (40 & 44) also did not change.
In fact I would safely say that there was no change at all that would indicate there was any control being exercised over the CPU temperature.0 -
-
-
I thought I had best update this thread.
I continued to seek explanations from HP personnel about the apparent problems I was having with the server.
I was passed from one foreign-sounding gentleman to another for weeks.
Eventually I received a communication from Ireland, and attempted to get the explanations I sought.
I asked the same questions over and over, phrased differently each time, and received answers which contained irrelevant information and no answers to the questions asked.
Eventually in all the verbosity emanating from HP there was a wee nugget .......... it appears I got an answer to my fan speed question, inadvertently.
Apparently there is a sensor which detects the presence of a card in the slot and sets a new minimum fan speed based on the presence of that card. As I have found out, it is not the actual physical presence of the card, but the fact that it is enabled that triggers this new fan speed.
At least I had my answer ....... HP set a new fan speed because an expansion card is plugged in and enabled, with NO reference to any contribution to temperatures anywhere within the case.
I accepted that as my answer although I could not but wonder why, after a period of time, the fan speeds were not controlled by temperature (Its not as if they do not have enough sensors in the rig!).
I continued to try to elicit WHY the temperature displayed in the iLO was a fixed value and unchanging for two of the sensors (the one I was really interested in was the CPU).
I received lots of BS, not of which addressed my question.
I continued to ask for a reason (politely I might add), until one day I was told "I am not going to pursue this any further with hp support."
From then on my emails were bounced.
End of!
I put the box aside for a few months ..... other things to do ...... but now have pulled it out again with a view to making use of it.
********
I will not be using a HP 'approved' OS on this box.
I have my own preferences ...... for instance if running as a HTPC I will likely use Openelec!
Before I do so I will be running a home distro on it for testing etc.
I have bought 3TB drives for it, so it will be working as a storage device regardless whatever else it does
It seems to me I will have to control the fans independent of any HP control system.
The implication from help received in this thread implies there is some form of 'hand-over' to the OS when using approved operating systems.
If anyone has any information on this aspect of things I would appreciate hearing it, thanks.
The alternative seems to be to disconnect the fans from the motherboard (put in a dummy load) and use a fan controller to manage things.
Thanks for reading.0
Advertisement