February 23, 20179 yr Can anyone tell me what's wrong? I am only getting 4mb/s with a wired connection and a cache drive. Edited February 24, 20179 yr by xhaloz
February 23, 20179 yr Your problem is this: Quote Feb 22 23:36:03 TheDeathStar kernel: qemu-system-x86: page allocation failure: order:5, mode:0x24040c0(GFP_KERNEL|__GFP_COMP) Feb 22 23:36:03 TheDeathStar kernel: CPU: 0 PID: 16371 Comm: qemu-system-x86 Not tainted 4.9.10-unRAID #1 Feb 22 23:36:03 TheDeathStar kernel: Hardware name: Supermicro X9SCL/X9SCM/X9SCL/X9SCM, BIOS 2.10 01/09/2014 Feb 22 23:36:03 TheDeathStar kernel: ffffc9000bfe7b50 ffffffff813a353e 0000000000000001 0000000000000000 Feb 22 23:36:03 TheDeathStar kernel: ffffc9000bfe7be0 ffffffff810cb40d 024040c000000040 ffffffff8193d22a Feb 22 23:36:03 TheDeathStar kernel: ffffc9000bfe7b78 0000000000000010 ffffc9000bfe7bf0 ffffc9000bfe7b90 Feb 22 23:36:03 TheDeathStar kernel: Call Trace: Feb 22 23:36:03 TheDeathStar kernel: [<ffffffff813a353e>] dump_stack+0x61/0x7e Feb 22 23:36:03 TheDeathStar kernel: [<ffffffff810cb40d>] warn_alloc+0x102/0x116 Feb 22 23:36:03 TheDeathStar kernel: [<ffffffff810cbedb>] __alloc_pages_nodemask+0xa59/0xc71 ... There are hundreds more like that in your syslog. Things I would try include a MemTest, because if you have bad RAM there's no point in continuing and the error does appear to be one of memory allocation. I'd look for a newer BIOS, and I'd try running without VMs for a while. In fact I'd shut down the VM service and see if that brings stability. Other than that, try a different kernel - either wait for 6.3.3 or roll back.
February 24, 20179 yr Author 2 hours ago, John_M said: Your problem is this: There are hundreds more like that in your syslog. Things I would try include a MemTest, because if you have bad RAM there's no point in continuing and the error does appear to be one of memory allocation. I'd look for a newer BIOS, and I'd try running without VMs for a while. In fact I'd shut down the VM service and see if that brings stability. Other than that, try a different kernel - either wait for 6.3.3 or roll back. So I am basically a noob when it comes to unraid. What exactly does this mean? How do I run a memtest ? My bios is from 2013 so yeah I'll update it. How will I know when the problem is fixed? Thank you!
February 24, 20179 yr I take it to mean that there's a problem with allocating memory and it seems to affect VMs or the hypervisor that supports VMs. It might be hardware related or it might be BIOS related or it might be a kernel bug. Sorry if that sounds a bit vague but it's the best I can offer for an explanation. To run a memory test reboot your server with a keyboard and monitor attached and select MemTest from the boot menu and let it hammer your RAM for a good long time like 48 hours. If the RAM fails then you've found a problem that you can easily deal with. If it passes then it gets more difficult to fix the problem. If you can find a newer BIOS it's worth updating. Shutting down the VM service gets rid of a lot of complexity so it can be a useful aid to troubleshooting. Since there are known bugs in the kernel trying a different kernel might help. When the problem is fixed you'll get the performance back that you're missing - did it happen after an update? - and your syslog will look a lot tidier.
February 24, 20179 yr Author 52 minutes ago, John_M said: I take it to mean that there's a problem with allocating memory and it seems to affect VMs or the hypervisor that supports VMs. It might be hardware related or it might be BIOS related or it might be a kernel bug. Sorry if that sounds a bit vague but it's the best I can offer for an explanation. To run a memory test reboot your server with a keyboard and monitor attached and select MemTest from the boot menu and let it hammer your RAM for a good long time like 48 hours. If the RAM fails then you've found a problem that you can easily deal with. If it passes then it gets more difficult to fix the problem. If you can find a newer BIOS it's worth updating. Shutting down the VM service gets rid of a lot of complexity so it can be a useful aid to troubleshooting. Since there are known bugs in the kernel trying a different kernel might help. When the problem is fixed you'll get the performance back that you're missing - did it happen after an update? - and your syslog will look a lot tidier. I have IPMI so I dont need to hook a monitor up. I am running the mem-test now I actually saw it upon a reboot attempt to update the bios. I just updated unraid a few days ago. I will try this mem-test and if it passes I'll do the bios update. I'm going to report back soon! Edited February 24, 20179 yr by xhaloz
February 24, 20179 yr Author I think I fixed it with the bios update. What do you think? Edited February 24, 20179 yr by xhaloz
February 24, 20179 yr Well, there are no page allocation failures logged in that syslog so you might think the problem is fixed. However, it's identical to the one you posted in your OP, alongside your diagnostics. So, I'm afraid I'm not entirely convinced. So use it for a while and then grab new diagnostics and we'll see.
February 24, 20179 yr Author 16 hours ago, John_M said: Well, there are no page allocation failures logged in that syslog so you might think the problem is fixed. However, it's identical to the one you posted in your OP, alongside your diagnostics. So, I'm afraid I'm not entirely convinced. So use it for a while and then grab new diagnostics and we'll see. So far I have not seen that error. I do want to note that I did have IOMMU in a disabled state, don't know if that has anything to do with it. Mem test passed for 12 hours. I have ECC Unbuffered ram that was expensive, I was hoping it wasn't failing.
February 24, 20179 yr Good. Can you live without IOMMU? Turning it off certainly removes a lot of complexity. I'm glad the new BIOS helped.
February 24, 20179 yr Author 34 minutes ago, John_M said: Good. Can you live without IOMMU? Turning it off certainly removes a lot of complexity. I'm glad the new BIOS helped. Well it was off before. I watched a video on youtube and it said I needed it (IOMMU). I have no idea why. But yeah I guess it was the BIOS. Hell I bought the supermicro mobo in 2014 with a 2013 bios image. So it was good that I updated to the most recent which was 2015. I mean I am not sure it fixed it but I do not see issues in my log.
February 24, 20179 yr You need IOMMU if you want to pass through hardware to VMs, like a a video card through to a Windows VM, for example. If you don't need that feature then it's OK. You can still run VMs that don't need hardware pass through, if you want to.
Archived
This topic is now archived and is closed to further replies.