Recommended Posts

I have the N36L version of this box. Love it to bits for playing about on. Running ESXi 5 and yes the slowest thing in this box really are the drives. I have three discs and try to stagger my VM's across them so no disc is really doing too much work.

I've have had 1 x 2008 R2 DC, 2 x Exchange 2010 running on 2008 R2, 1 x Windows 7 and a Arch Linux. They run well as long as each one isnt under too much load (Disc I/O).

Anyone managed to get OSX working? Works fine on a machine in my office running an Intel CPU but on this Microserver it just sits with a grey apple logo. I figure its the AMD CPU and pretty much given up trying now.

Here is the ram I ordered -- cheaper, and have had not issues with it.

20-148-347 MEM 4Gx2 | CRUC CT2KIT51264BA1339 R $42.99

http://www.newegg.co...N82E16820148347

Looks like the price went up a couple of bucks.

Thanks! I can install vSphere 5 then now. Got it from the UK site however!

I just have to serve up the files via smb.. I don't have to transcode them on the fly or anything.. Just a simple file server, my popcorn hour plays the files in whatever format they are in.

So you were writing 3 different installs to the same datastore at the same time - and reading all from the same iso?? WTF yeah that going to be slow as hell! Why not just install once and then copy the install if you were installing the same OS?

Yeah, I used to have a Popcorn Hour but didn't like the UI. Will be migrating everything over to Raspberry Pi with XBMC/Plex shortly anyway. That said, I've not used a Popcorn Hour since 2008. I'm guessing they've improved somewhat since then? All I remember about it was power cycling it about 50 times a day.

I would think the NL40 is quite limited in regards to performance, it would only take a number of VM's for it to choke.

Well, the only VM thats really doing any work is the pfSense VM and the Sabnzbd/Sick Beard VM, running 6 total.

"it would only take a number of VM's for it to choke."

No **** ;) But you can say that for any hardware - the number of VMs you can run at the same time is just going to vary. And what those VM's are doing.

What I can dell you is it runs my router, and is running my file server, and my test box w7 and my linux distro. The file server, the router and the linux box are on all the time..

It is more than capable of handling the number of VMs I would ever need to be running at the same time, and allow for many many more to be available when I need them for play or testing. Allowed me to combine my file server and my router boxes into a tiny little thing - that looking currently at the killawatt meter its connected to is drawing 55.5 watts ;) That is with 4 disks in it. Its been on for 366 hours and has drawn a total of 20.3KWH, so I pay like 12.5 cents per kwh so your looking at $2.50 to operate for 15 days. So about $5 a month to have it on running all the machines I need to run or play with.

And for the price you sure and the hell can not beat it for a home esxi solution or even a small business setup. No I wouldn't expect to run my enterprise VMs on it ;)

Looks like I found a pretty big part of what was happening with the I/O problems too.

575225_345153122214439_100001592618629_987075_220579141_n.jpg

See the Disk I/O. Notice the drop, thats because one of my Hard Drives had some damaged sectors and ESXi wasn't marking them as bad on the fly, causing massive I/O and Error Rates, after yanking the drive it dropped massively and all VMs are running great now.

SK[' timestamp=1335380702' post='594824025]

I have the N36L version of this box.

I have a N36L as my primary storage server, Windows Server 2008 R2, Software RAID5 across 4 Drives (2TB). 5.5TB Total Storage.

SK[' timestamp=1335380702' post='594824025]

Anyone managed to get OSX working? Works fine on a machine in my office running an Intel CPU but on this Microserver it just sits with a grey apple logo. I figure its the AMD CPU and pretty much given up trying now.

OS X runs fine with ESXi on a XServe / Mac Pro. But it doesn't even want to attempt a startup on this HP Hardware. Would require a hacked Kernel I think, and I wouldn't touch that with a bargepole!

If anyone needs discounts on any vmware stuff I can help since I am a partner.

Nice offer! But ESXi is Free. That said I will likely need some Workstation Licences in the coming months so I may touch base with you on that at some point.

Just as a further update to this thread.

My VSwitch is configured to run at Full Gigabit, MTU Set at 9000. But I am unable to get more than 100Mbit out of the VMs, see below.

post-180771-0-06510000-1335431460.png

ideas?

As you can also see, the individual NICs are set to Gigabit speeds also.

post-180771-0-92438700-1335431593.png

Why do you have your mtu at 9000? Do you have jumbo frames setup on your nics? Does you physical switch support jumbo?

BTW - what is producing that stats page? Showing your disk io and network and memory?

Your graph is terrible why is it all over the place? See mine that is from the VM, and that copy dialog is from my workstation. Also notice my VM nic is showing connected at 10GB, not just 1 ;) I never changed the mtu from default 1500 -- there is NO reason too. Does every device on your network have a 9000 mtu? If not your switch is doing a **** load of work.

post-14624-0-09917400-1335438634_thumb.j

From a VM transferring a file to the host ESXi server via SCP I get about 2MB/s... I'm not even kidding, it's abismal.

From one VM host to another on a private internal 'network' I get about 18MB/s.

That's on a dual quare core 8GB server with 2 VMs

SK[' timestamp=1335438001' post='594825411]

I get the same problem, to be honest I haven't looked into it yet, would be interested if you find a fix. it was doing it over iSCSI from a Windows 7 to FreeNAS VM for me.

Keep an eye on this thread then.

Why do you have your mtu at 9000? Do you have jumbo frames setup on your nics? Does you physical switch support jumbo?

BTW - what is producing that stats page? Showing your disk io and network and memory?

Your graph is terrible why is it all over the place? See mine that is from the VM, and that copy dialog is from my workstation. Also notice my VM nic is showing connected at 10GB, not just 1 ;) I never changed the mtu from default 1500 -- there is NO reason too. Does every device on your network have a 9000 mtu? If not your switch is doing a **** load of work.

I changed up my MTU to 9000 to see if that helped increase the speed, and it did, by 3MB/s. So I've lowered it back down.

How on earth did you get a 10Gbit to show up in ESXi?

Regarding the Graph. Thats VMWares Client for the iPad, the graphs were all over the place until I ripped out that dodgy hard drive, now they're holding stable with no issues.

From a VM transferring a file to the host ESXi server via SCP I get about 2MB/s... I'm not even kidding, it's abismal.

Yeah, transferring directly to ESXi gives me horrid performance too, took about 3 hours to transfer my Server 2008 r2 image over the Console, and about an hour and a half using SFTP.

SK[' timestamp=1335438801' post='594825437]

I know when I tried to set my box to 9000 it said the nic didnt support it.

My entire network supports 9000, I just don't use it normally, this was the one case where I did.

10GB is autoneg with driver for vmxnet3

here

post-14624-0-85388900-1335439344.jpg

yeah putting files on the datastore directly blows chunks.. I found a lot of threads about ftp server on the older 4.1 esxi that helped, but have not found good thread about putting ftp on esxi 5 yet.

But really only files you have to put there are iso for your installs and patches/upgrades to esxi -- so not too worried about slow speed uploading files to the datastore - not something have to do very often.

It would be supported by anything that you can install the tools too - this has the drivers for it. I had them running on my pfsense box (freebsd 8.3) until I ran into a weird issue with connecting to ipsec/l2tp vpn from behind my router.

Freaking strangest thing -- I have to spend some time documenting it so can put in a bug report. But I could just not get a connection, it would just hang - I even turn on my oakley.log in windows to see the details of the ipsec connection, and nothing looked wrong - it just would hang. As soon as I put the nics back to e1000 on my pfsense vm - bam connected right into my work vpn from my phsycial laptop.

Let me grab a large file from vmfs, my transfers are all too and from RDM drives.. I assume yours are from a vmfs disk?

Freaking strangest thing -- I have to spend some time documenting it so can put in a bug report. But I could just not get a connection, it would just hang - I even turn on my oakley.log in windows to see the details of the ipsec connection, and nothing looked wrong - it just would hang. As soon as I put the nics back to e1000 on my pfsense vm - bam connected right into my work vpn from my phsycial laptop.

Yeah I decided against using the vmxnet3 drivers on FreeBSD for the time being.

Ok that is pretty bad! This is writing to my vmfs ? of my vm..

post-14624-0-32268800-1335440223.jpg

Now I run my system drives on my vms as thin provisioned.. So I am quite sure its having to expand out the file system as I copy that file. Let me do a copy back after it gets up there, and I can test read speed from it and then writing file when the file system has already been expanded out, etc.

Ok this copy file from vmfs -- not as good as the RDM drive, and it took a while to ramp up see the networking graph. But its workable I guess. But RDM drive is much better ;) Maybe that is your issue. Give the vm raw access to the disk vs a vmfs.

post-14624-0-41445000-1335440585_thumb.j

Well thats not too shaby there.. I deleted that 3gb file I had copied up to the vmfs, so it would of expanded the drive. Then copied up a different iso (backtrack) and 70MBps -- can live with that for sure.

post-14624-0-62670100-1335440785_thumb.j

Well, thats pretty terrible, and the same kinda speeds I've been seeing.

The Server 2003 R2 box I've setup for sabnzbd was choking, but as the file system has been expanded it's got better.

The Problem comes in as

Incomplete Downloads go to a eager zeroed VMFS

Once downloaded they expand on this VMFS

then move to a Downloads_Complete folder located on a N36L with a dedicated Windows Server 2008 R2 OS

Finally, Sickbeard picks up the file and moves this to a Media Folder after doing some processing on the files.

Annoyingly, I can download a 300MB file faster than I can move it across my network in some instances.

Ok I did copy back from that vmfs, and its rocking.. I even rebooted my workstation to make sure nothing was cached since I had just copied up that iso of backtrack, etc..

post-14624-0-15947300-1335442094.jpg

Can you do a simple iperf test to your vm.. is the bottle neck network or disk??

So as you can see, I see between 750 and 800Mbps with iperf to my vm from my workstation. What do you get?

post-14624-0-09116700-1335442159_thumb.j

Just simple math of 750/8 tells me I should at max there be able to copy at 93MBps -- i am seeing 70 to 80.. So works out pretty well. And when I use that nastester I do believe its close to 90 even.

This topic is now closed to further replies.
  • Posts

    • Waymo recalls self-driving software after cars enter closed freeway work zones by Paul Hill Waymo, the self-driving car maker owned by Alphabet – the parent company of Google –, has recalled some of its fifth-generation Automated Driving Systems (ADS). It did so after some of its cars drove through closed construction zones. According to the National Highway Traffic Safety Administration (NHTSA), the affected vehicles were capable of driving through a closed freeway construction zone and continuing to drive at speed. The listing on the NHTSA website says that Waymo is currently developing a solution to fix this issue, but in the meantime, freeway driving is being restricted. Waymo will update its ADS software so that vehicles can detect when they can avoid entering construction zones. According to the Safety Recall Report, on April 20, 2026, Waymo’s Field Safety Committee began meetings reviewing an event from April 11, 2026, and five events from April 19, 2026, where Waymo’s autonomous vehicles didn’t recognize and drove past ramp closure signs into the pre-planned freeway construction zones. This took place in Phoenix, Arizona. Separately, on May 18, 2026, seven Waymo vehicles entered freeway lanes with active construction in the San Francisco Bay Area by driving between cones that were placed to show the lane was closed. On the back of both of these events, Waymo restricted freeway driving until it could address the issue. In June, Waymo’s Safety Board reviewed the issue and additional information related to ADS performances around construction zones; then, as a result, it decided to conduct a recall. This development is not good for Waymo as it adds to a growing list of technical hiccups its cars have experienced. Ultimately, it will lead to more scrutiny from lawmakers around the world who will be more cautious about letting autonomous vehicles on their roads without tighter regulation. For readers in areas where Waymo operates, does this news make you more wary about stepping into one of these vehicles?
    • I'm still on Windows 10 22H2 because I didn't want to deal with all the issues in Windows 11, so I waited almost a week before installing the latest Patch Tuesday update (KB5094127), I went ahead and did it, and it was a huge mistake—ever since then, my File Explorer has seen a performance drop of about 30% when transferring large files... Once again, Microsoft has outdone itself! This update cannot be uninstalled, either through the Control Panel (via Settings) or by accessing Advanced Startup Options. The only possible alternative would be to use system restore points, but I’d have to reinstall all app and driver updates (and there’s no guarantee it would work). Or there’s the “nuclear option” of a in-place repair without losing files or apps, but even then, all my customizations would be lost! Microsoft just can’t help but mess everything up! Way to go, Microsoft! But I still don’t want your c****y Windows 11!
    • Microsoft: Windows 11 could finally solve a major issue across AMD, Nvidia, and Intel GPUs by Sayan Sen While Microsoft has been trying to improve it, Windows 11 is definitely not flawless, as even today some issues are taking a year to publicly acknowledge. However, one area of trouble that may finally see much better results soon is graphics driver crashes. Work on graphics driver timeouts, also called Timeout and Detection Recovery (TDR), is not new as the latest WDDM 3.2 also has specific improvements regarding it. Windows Display Driver Model (WDDM) version 3.2 is supported on Windows 11 24H2 and 25H2. However, with the upcoming version 26H2, TDR crash diagnosis could go to the next level as Microsoft is introducing a new DirectX 12 API feature called "DirectX Dump Files". Similar to how system memory dump files work when a system crashes or freezes or encounters any such major issue, DirectX Dump Files (DDF) will essentially record a snapshot of the GPU execution right at the moment a graphics-related crash or hang or freeze occurs, so that developers can better understand and diagnoze these TDR and timeout detection errors. The dump will be available as a .dxdmp file for analysis and it will be a comprehensive dump file generated with detailed insights about the hardware, drivers, Windows, as well as the affected application. This should be another welcome change in this department. Earlier at GDC 2026, when the technology was first debuted, Microsoft had shared more details regarding it. The company had explained how DDF is designed to gather data from every layer of the graphics stack into a single file, eliminating the need for developers to manually correlate logs from multiple tools. As mentioned above, the dump can contain a lot of useful details like GPU hardware state information such as register values, shader program counters, page fault virtual addresses, shader memory data, and command buffers. Alongside that, it also captures DirectX runtime and kernel information, including D3D objects, pipeline state objects, device error data, adapter details, and CPU call stacks. Microsoft says the feature has been built around two primary use cases: retail device removals and local device removals. The former allows developers to collect crash information from end users' systems in the field, while the latter helps QA teams and developers investigate issues on test machines. Developers will also be able to include up to 2 MB of custom application data through new D3D12 APIs, providing additional context for troubleshooting. In addition, Microsoft is introducing three dump collection modes ranging from zero-overhead capture, which has no runtime performance impact on supported hardware, to higher-detail modes that collect more vendor-specific debugging data. On compatible Tier 2 hardware, zero-overhead dumps will be enabled by default, meaning developers may begin receiving useful crash diagnostics without making any code changes. The table below explains the three tiers: Tier Description NO_OVERHEAD Enables crash capture with no runtime cost and is suitable for broad deployment MEDIUM_OVERHEAD Provides a balance, capturing additional diagnostic data with moderate impact HIGH_OVERHEAD Collects the most detailed GPU and driver state available, enabling deeper investigation at the cost of higher runtime overhead In terms of availability, the company expects broader release to be around the fall of 2026, which should be right around the time when Windows 11 version 26H2 lands. Right now, DirectX Dump Files are available as a preview and currently, only AMD has the compatible AgilitySDK Developer Preview driver version 26.10.07.02. You can find the official announcement post here on Microsoft's website.
    • And with SO much better perf than the laggy mess that is Files.
  • Recent Achievements

    • First Post
      BizSAR earned a badge
      First Post
    • Week One Done
      Jordan Smith earned a badge
      Week One Done
    • Reacting Well
      BizSAR earned a badge
      Reacting Well
    • First Post
      AndreaB earned a badge
      First Post
    • Week One Done
      Huge Trailer earned a badge
      Week One Done
  • Popular Contributors

    1. 1
      +primortal
      598
    2. 2
      +Edouard
      190
    3. 3
      PsYcHoKiLLa
      80
    4. 4
      Michael Scrip
      76
    5. 5
      Steven P.
      69
  • Tell a friend

    Love Neowin? Tell a friend!