ehfortin

Hi, I'm having 3 of these for my ESX cluster. 2 based on the i3-4150 and 1 that runs on a E3-1220. To answer the question, the fans can be loud. They will be while booting which can take a few minutes as HP is checking a lot of stuff behind the scene. Then, the fans will slow down and adjust to the temperature inside the casing. So if your room is warm and you use a lot the CPU, the fan will accelerate up to what is needed to keep everything cool. Adding CNA, HBA or video card can also increase the internal temperature a lot which will translate in more noise. Just to give you an idea, my 3 machines are in another room with a closed door. I can hear the fans accelerate when I'm loading the CPU. Note that it is not a high pitch noise like small 1U/2U servers as the casing and fans are a lot larger. If the machine is in the same room as you, you will definitely hear it. That being said, these are really good machine for lab usage. The ILO is giving you remote access to the console which is great. However, the "basic" is only for management. It will stop acting as a remote KVM about 30 seconds after the OS start booting. You can buy a license that gives you the ability to use that without limit. It is a a few hundred $$ but you can find original licenses on Ebay for a lot less. Then, you have a nice remotely manageable server that you can control even if the server is powered off, as long as there is electricity connected to the power supply. That's very useful when you have your servers in another room ehfortin

Great! Thank you Tom!

If that can help to figure what I'm talking about, the pvscsi is the switch "CONFIG_VMWARE_PVSCSI" in the .config file. For anybody that want to compile this, that is under Device Drivers / Scsi device support / SCSI low-level drivers / VMware PVSCSI driver support when doing a "make menuconfig". Thank you. ehfortin

Hi Tom, I can see in the release note for RC4 that you added SCSI generic support (sg). I don't know if it was related to my request but what I was asking for is the support of pvscsi which is the switch "CONFIG_VMWARE_PVSCSI" in the .config file. That is under Device Drivers - Scsi device support - SCSI low-level drivers - VMware PVSCSI driver support when doing a "make menuconfig". It would be great to have this built-in in the next RC or final. Thank you. ehfortin

Hi, I know. But doing the passthrough force you to have a motherboard/chipset/cpu that allow this and they are not that common. You basically need component done for servers. I was not willing to invest that much money to run Unraid as a VM. So I'm using RDM and I've compiled a kernel with PVSCSI into it and it allow me to let VMware ESX manage the disks (and the hot-plug stuff that is supported with my LSI card) and to get maximum performance at the guest level. Let say it is just another alternative that is cheaper and only cost a recompile of the kernel with the appropriate switch enabled. If the information is useful to somebody, that's the goal. ehfortin

Hi, As for VMXNET which has been available for a long time in the Linux kernel, there is the equivalent named PVSCSI which give us the opportunity to use the "paravirtual" disk driver under VMware ESX which increase the performance under ESX (the paravirtual driver allow the guest OS to redirect the call to the ESX SCSI driver so we eliminate a layer which is always good performance wise). This has no impact on any other disks drivers in the kernel. Note that it has to be staticly linked (not a module) if you want to be able to boot on a disk with the paravirtual driver. Thank you.

Thank you BRiT to summarize those changes. It make our life easier ehfortin

Found the problem. If you someone has the issue where Unraid is showing sd drives but no mention about the serial number of the disk (which make it impossible to add in Unraid), you just have to use "vmkfstools -z /vmfs/devices/disk/vml.xxxx diskname.vmdk -a lsilogic" while creating the RDM instead of the "- r" switch that is documented in the official procedure that is published at http://lime-technology.com/forum/index.php?topic=14695.0 Hope this is useful. ehfortin

Hi, You are right. The question has been prompted by a remark that was related to RC2 but I should I moved the topic elsewhere after the initial answer. Have a nice day. ehfortin

Hi all, I tried to do some search on this now lenghty topic but without success. I've installed Unraid under ESXi 5. I have an LSI 1068E based card which is recognized by VMware. However, as I don't have a VT-d supported board, I have to use RDM. So, I created my RDM and assign those to the Unraid VM. After booting in Unraid, I can see all the drives but without much details (I only see the drive name and the size). This is very different from the complete identification I'm getting when booting directly Unraid on bare metal. My problem is that I have identified all my drives correctly and know exactly which sd device is which physical drive but when I select one of those in the select box in Unraid (those under the column "Identification"), it reload the page and I'm back to the "no device" selection. At first, I was thinking it was related to problem with /dev/disk/by-path but 1) apparently it is not used anymore (now using /sys/dev/block) and 2) I can see my drives on both place anyway. I've looked at the log and there is nothing. Each time I select a device, the log refresh with the following: May 8 12:05:57 fileserver kernel: md: unRAID driver removed May 8 12:05:57 fileserver kernel: md: unRAID driver 2.1.3 installed May 8 12:05:57 fileserver kernel: mdcmd (1): import 0 0,0 May 8 12:05:57 fileserver kernel: md: disk0 missing May 8 12:05:57 fileserver kernel: mdcmd (2): import 1 0,0 May 8 12:05:57 fileserver kernel: md: disk1 missing May 8 12:05:57 fileserver kernel: mdcmd (3): import 2 0,0 May 8 12:05:57 fileserver kernel: md: disk2 missing May 8 12:05:57 fileserver kernel: mdcmd (4): import 3 0,0 May 8 12:05:57 fileserver kernel: md: disk3 missing May 8 12:05:57 fileserver kernel: mdcmd (5): import 4 0,0 May 8 12:05:57 fileserver kernel: md: disk4 missing May 8 12:05:57 fileserver kernel: mdcmd (6): import 5 0,0 May 8 12:05:57 fileserver kernel: md: disk5 missing May 8 12:05:57 fileserver kernel: mdcmd (7): import 6 0,0 May 8 12:05:57 fileserver kernel: md: disk6 missing May 8 12:05:57 fileserver kernel: mdcmd (: import 7 0,0 May 8 12:05:57 fileserver kernel: md: disk7 missing May 8 12:05:57 fileserver kernel: mdcmd (9): import 8 0,0 May 8 12:05:57 fileserver kernel: md: disk8 missing May 8 12:05:57 fileserver kernel: mdcmd (10): import 9 0,0 May 8 12:05:57 fileserver kernel: md: disk9 missing May 8 12:05:57 fileserver kernel: mdcmd (11): import 10 0,0 May 8 12:05:57 fileserver kernel: mdcmd (12): import 11 0,0 May 8 12:05:57 fileserver kernel: mdcmd (13): import 12 0,0 May 8 12:05:57 fileserver kernel: mdcmd (14): import 13 0,0 May 8 12:05:57 fileserver kernel: mdcmd (15): import 14 0,0 May 8 12:05:57 fileserver kernel: mdcmd (16): import 15 0,0 May 8 12:05:57 fileserver kernel: mdcmd (17): import 16 0,0 May 8 12:05:57 fileserver kernel: mdcmd (18): import 17 0,0 May 8 12:05:57 fileserver kernel: mdcmd (19): import 18 0,0 May 8 12:05:57 fileserver kernel: mdcmd (20): import 19 0,0 May 8 12:05:57 fileserver kernel: mdcmd (21): import 20 0,0 Doesn't say much except that it doesn't recognize the disk I'm selecting. Even if I use a VMDK that I assign as a drive, I get the same result. Is this something that has been reported before? Is there a known fix? Thank you. ehfortin

Hi, Thank you. I found it in the documentation as well. What I didn't found however is how to recreate a bzroot from what have been compiled and tested in Slackware (I'm using Salix 13.37 as Slackware still doesn't answer...)? Do you have this documented somewhere? Thanks. ehfortin

Hi BRiT, *** Update *** I just found updated documentation for RC2 at http://lime-technology.com/wiki/index.php?title=Installing_unRAID_5.0_on_a_full_Slackware_Distro. I just have to find a way to get the latest version of slackware and I'll try this. **** I've installed 5.0-rc2 on ESXi 5.0 on a board that don't support VT-d. As such, I use RDM to pass my disks to the VM. This used to work fine back on 5.0-b2 but if I remember well, I had to recompile the kernel with the VMware paravirtual driver. Right now, using the LSI SAS driver in ESXi, I'm seeing the drive in Unraid but without any label/path in id-path/id-label. As such, Unraid seems unable to make sure the drive are the same so I can't assign them. Using the paravirtual driver, I see no drives which is not good. Following your post on this: I've looked at the wiki on this subject and the steps to compile a new kernel use Slackware as the basis. The problem I have now are 1) Slackware website is not responding at all and 2) the latest and greatest kernel in Slackware is 2.6.xx which is far from 3.0.30. Can you help me (and probably others) about how you proceed to compile this latest release of Unraid at this moment? Is there a simple recipe to follow that is easier then modifying at each step the procedure on the wiki? Thank you. ehfortin

Hi, I'm having a SuperMicro card which is actually an LSI 1068 chipset. As it use the mptsas driver, I assume I would have the same issue going on B13,B14 and RC1, right? Is the custom kernel you made something you can distribute and that we can replace easily on the USB key compared to what is installed by default with B14/RC1? If it is doable, I would prefer this option then trying to create my own kernel. I've done that a few years ago and it worked (I actually installed unraid on Salix at that time) but... I remember it was a long process and I don't have the time for this anymore. Let me know if this is possible to distribute at all or if we really have to create our own kernel. Thank you. ehfortin

Yes, it would be a good idea to not try to autostart the array when there is a version change. By displaying the release notes (or something like it) and showing what emhttp think is appropriate but letting people decide to start the array or not based on these info, it may reduce problem. At least, people that sees errors and are not too sure what to do could ask question before trying it and/or revert back to the original version until they are more confident with the new version. ehfortin

Hi, It took some time to figure that the script unraid_partition_disk has been updated between the time I've downloaded it yesterday morning and ... later the same day. I was unable to get success with the original one. Now, I have an interesting situation with my cache drive. At first, I figure it was not a problem as I knew it was empty so I tried just to reformat the disk with Unraid. Well... It stays in the unformatted state. This morning, I've tried to use Joe's script with apparent success (the script did create a MBR that is Unraid compliant). I've done the reiserfsck and everything whas fine. I'm even able to mount it and see that it is empty. However, Unraid continue to say it is unformatted. Here is the extract from the script: root@fileserver:/boot# ./unraid_partition_disk.sh /dev/sdb ######################################################################## Model Family: Seagate Barracuda 7200.10 family Device Model: ST3250410AS Serial Number: 9RY1CT09 Firmware Version: 3.AAC User Capacity: 250,059,350,016 bytes Disk /dev/sdb: 250.1 GB, 250059350016 bytes 1 heads, 63 sectors/track, 7752336 cylinders, total 488397168 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000 Device Boot Start End Blocks Id System /dev/sdb1 63 488397167 244198552+ 83 Linux Partition 1 does not end on cylinder boundary. ######################################################################## ============================================================================ == == DISK /dev/sdb IS partitioned for unRAID properly == expected start = 63, actual start = 63 == expected size = 488397105, actual size = 488397105 == ============================================================================ I've retried to format it from Unraid and, after saying it is formatting, it still shows unformatted, even if I stop and start the array or if I reboot. I'm including the syslog. However, I'm pretty sure I've identified why the problem is occuring but don't know how to fix it. When I look in the log (and if I try to mount it manually), I get the error "Mar 2 09:00:23 fileserver logger: mount: unknown filesystem type 'ddf_raid_member' Manually I can force the type to be reiserfs and it works fine. It seems like Unraid is doing a mount without telling the type is reiserfs so... it get the default ddf_raid_member. Who knows how to change that? In fdisk, I can see the partition ID is 83 (Linux) like all others. But that doesn't relate to actual file system on the disk as Linux type is working with ext2, ext3, ext4, reiserfs, etc. I also have another question. The default in Unraid 5b6 is to format MBR on 4K aligned. What if I put an older 1 TB drive that was not in that new format? Does it work or I have to change the default format to MBR unaligned for it to work? Thank you. ehfortin zipfile.zip