Mobo and other HW suggestions wanted for new OMV box with ECC

  • Yeah I'm aware of the global shipping programme. I'll take another look once all is up and running. I haven't heard of Matrix Orbital. I'll have a look at those too.



    I finally got some time this evening to start assembling. It's looking good ;)


    The default way Thermaltake had run all the wires was crappy - typical quick factory job. I think I spent more time re-routing wires than anything else.


    I wouldn't worry about it looking like a viper nest... I think I've gone through almost 30 cable ties. >:D



    I still have 9 SATA cables to do so that'll probably be closer to 35.


    I ordered those today but they're coming from HK and probably won't get dispatched until Monday so I won't be up and running yet. I didn't want to get SATA cables until everything was in and I'd checked lengths. I have 9 of them to route, so didn't want to guess. I couldn't find those braided ones you mentioned :( however, I managed to find some rounded black/dark grey ones for the array and a flat blue one for the OS SSD. They're all about 50cm because for some reason 40cm cables are more expensive... Probably because you 'need' them.


    Here's a view through the rear vent...



    Btw.. It's VERY quiet. I can't imagine the drives will make a lot of difference. They're reasonably quiet models.


    After a couple of preliminary benchmarks, and letting everything settle again, most components are sitting in the mid 30s or early 40s C, with the exception of the CPU which sits at about 53C. I might lap the heat sink when I get an hour. I think that can come down a tiny bit more.

  • The SATA cables? Yeah it's going to be fun, but cable ties will prevail :D or do you mean lapping the heat sink? My least favourite job... It's so mind numbingly boring which is why I haven't done it yet. Lol.


    I tested the box at idle this morning - obviously this is without the Seagate drives in it, but it sits at 86w (0.09kWh). That works out at about £10 a month if I keep it on 24/7 which I won't be doing. Pretty pleased with that. :)

  • Btw.. It's VERY quiet. I can't imagine the drives will make a lot of difference. They're reasonably quiet models.


    After a couple of preliminary benchmarks, and letting everything settle again, most components are sitting in the mid 30s or early 40s C, with the exception of the CPU which sits at about 53C. I might lap the heat sink when I get an hour. I think that can come down a tiny bit more.

    I'm half sure that Intel is still using that piece of ... thermal paste between the die and the heat spreader.
    While I'm not certainly advising delidding that baby, I think that it will bring more benefit than lapping the heatspreader.


    Not that matters much anyway. I'd check all CPU and powersave options in BIOS first.


    Also. Large fans are ninja. lot of air low noise. noise scales with rpm after all.

  • I don't think I want to risk disassembling the CPU. That 'could' be the quickest way I'll waste £350. I don't think 53C is too bad. It's just not as low as I would like. I also have to bear in mind that as the only pre-owned part in the system, I have no warranty with it - not that I would if I disassembled it anyway. When benchmarking the temps were reasonable (about 72C) then quickly dropped back down due to the water cooling unit.

  • @bobafetthotmail


    Might have some answers as to why it's 'running hot'... I was monitoring the temps in IPMI.. I think it's wrong.


    Here is a photo of the temps in IPMI:



    Now look at what the OS reports:



    This is what I would expect to see. Looks like a good 20c difference :-S is IPMI showing a mean temp or something?


    That would explain why the fans didn't spin up much while benchmarking. Probably didn't get above 55c. I'll test again today and post results.

  • Temps while geekbenching according to the OS:



    IPMI reported 58c when I caught it on a refresh.


    Btw, I am aware all cores aren't shown in the sensor readings. The OS is seeing all the cores/threads, but I think 13 must be the limit for this applet.

  • Might need an IPMI firmware update.


    IPMI is not the board's BIOS, it has its own firmware. It's a weak ARM SoC soldered to the board, talking directly with BIOS or running the board's ethernet when the server is powered down (which is why you can control everything and reboot and do whatever).
    It's basically your own little hardware backdoor lol.


    Here, (if this is the correct board) click on "IPMI firmware", see if version is higher, then do wat they say. http://www.supermicro.com/prod…rd/Xeon/C600/X10SRA-F.cfm


    If for some reason after the update you cannot log in again in IPMI (while the motherboard works fine) drop a mail to Supermicro support and they will provide a tool to unlock it, or look around as there are tools and instructions to do it on your own (being standards-compliant and all helps this).
    Also some IPMI updates on Asrock boards had similar issues (and similar quick answers by support with a unlock tool).

  • Hmm. It's up to date.


    I know it's a different chip and has its own firmware but I would have thought that it would get its readings from the BIOS and have them displayed correctly. This seems like a MASSIVE design flaw to me. Especially if one is relying on IPMI to monitor sensors. At least I know I can probably ignore it and get readings from the OS instead.

  • Just an update: On my Mobo you need to move jumper JPB1 to see the hardware temps in the BIOS instead of IPMI (this also disables IPMI for some reason).


    I did this, to confirm what temps the actual BIOS sees:



    You'll notice this is almost the same as physical ID 0 at idle.


    This is definitely a problem with IPMI.


    I'll drop Supermicro and email and put up with it short term. I don't really want to disable IPMI and this is the only issue I can see with it. No doubt, this will be patched in a future release. The main thing is that the BIOS sees the correct temps (and the OS) and I can use those for monitoring. As long as the BIOS sees the correct temps, the fans will spin up accordingly - and as previously mentioned, they don't really spin up at all. Although if the load temps don't exceed ~46C, this would explain why.

  • Zitat

    jumper JPB1 to see the hardware temps in the BIOS instead of IPMI (this also disables IPMI for some reason).

    That's what that jumper does. It disables BMC hardware (baseboard management controllers), like IPMI and the Intel management thing I forgot the name of....
    http://www.supermicro.com/support/faqs/faq.cfm?faq=18001


    Zitat

    I know it's a different chip and has its own firmware but I would have thought that it would get its readings from the BIOS and have them displayed correctly. This seems like a MASSIVE design flaw to me.

    either IPMI or BIOS bug.

  • Have been searching on Google. Looks like IPMI from Supermicro and other vendors have a history of problems. I googled "IPMI wrong temp" and got a ton of results. Some of them are a couple of years old, but this is not an isolated incident.


    I've got the number for Supermicro UK. I'm going to call them tomorrow and see what they say. A board of this price should not have these kinds of errors.


    I hope it's not back to the drawing board to find another mobo. :(

  • Right.. Another update.


    Been on the phone to Supermicro this morning and was hoping that there may be a BIOS or IPMI update that was pending for release but sadly there isn't.


    I've been advised to download Super Doctor 5 to try and narrow down the problem - if I can get it to download... It's coming in a 62K/sec and keeps failing at ~172MB and there are no offered mirrors.


    I've also just noticed another error in the IPMI:



    It shows that the memory is in A2 and B2 - when they are in fact in A1 and B1 as per the recommended placement in the included quick start and the manual. I've emailed again to ask for clarification on which is correct.


    I'm guessing other Supermicro users here haven't experienced anything like this? I'm starting to wonder if I'll be RMAing this board and looking elsewhere. There are a lot of warning signs here.

    • Offizieller Beitrag

    My X10SL7-F works very well. DIMMs are correct and cpu temp matches coretemp.

  • :-S I don't get what's going on with my X10SRA-F then. I know it probably doesn't mean too much as it's probably board specific, but what version of the IPMI firmware are you running? 1.77? BIOS 1.0a?


    The guy on the phone today has been very helpful, but I do get the impression that he and the other tech guys are scratching their heads a bit. I'm waiting to hear back from the guy tomorrow (I think he's emailed the U.S. office for input) with another download link for Super Doctor and some links to previous IPMI firmware and BIOS revisions to see if that makes any difference.


    I hate to add another problem to the list, but I also noticed this evening that clicking on the Expert Mode icon at the top of the BIOS screen doesn't appear to reveal any advanced options at all under any of the sections. I double checked to see if this was because of a jumper or other setting but I cannot get any extra options to appear.


    I'm not sure if this is a buggy IPMI/BIOS or if the board itself is damaged - it doesn't appear to be malfunctioning in any other way so I really hope this is just a simple bug fix needed - but I'll be honest... I'm concerned. It seems unlikely that any company that comes to highly recommended would release a product with some pretty big flaws in both the boards BIOS AND IPMI firmware.


    This is my first Supermicro purchase and:


    - IPMI reports wrong temps
    - IPMI identifies wrong memory banks
    - expert mode in BIOS doesn't work


    Also:


    - Supermicro's site clearly has some speed/timeout issues and no mirrors.


    It's not that I am completely writing this off (yet) but it's not filling me with confidence. This board cost almost £300.. It shouldn't be doing any of these things. Maybe one... But not all 3. Things are all starting to look a little cheap and rushed.


    Just in case this can't be resolved by the tech guys, I'm going to start looking for a replacement. In the event that this isn't fixed, I want to have a replacement in mind to have it ordered here quickly if the Supermicro does need to be returned.


    I've got my eye on a Gigabyte MU70-SU0. The only problem with this board is that it uses a narrow ILM.. I don't think the H105 will fit that. Bummer.


    I'll have to see what happens tomorrow. Hopefully this is all resolved as quick as I can click the flash button. ;)

    • Offizieller Beitrag

    I have had five supermicro boards - 3 x H8QGL-IF (quad G34 processor), X10SLV-Q (no IPMI), and X10SL7-F. They have all been flawless and IPMI worked perfect on the boards that have it. I did notice on an Amazon review of your board that someone was going to RMA it.


    Versions in attached pic.

  • Thanks for the screenshot.


    Did they say why they were RMAing it?


    Everything else has been good. I'm willing to try the same board again if this one has to go back. It's worth a second shot, but I think this is all firmware related.


    I called Supermicro again today for an update as I hadn't received an email with the Super Doctor link or any other info and the guy said he was waiting to hear back from their firmware guys to see if there was a pending update that I could have early. He hoped it would be by close of play today but I haven't heard anything.


    Watch this space.

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!