typewriter

Members
  • Posts

    48
  • Joined

  • Last visited

Everything posted by typewriter

  1. I want to upgrade my Unraid Server with a modern HBA which is also capable to work with NVME (U3) and PCIE 4. Those are expensive but should work for a longer timespan. I found this one: Broadcom 9500-16i HBA 16-Port SAS/SATA/NVMe Is this one fine or does anyone have a better suggestion?
  2. Asustore even made own videos how to install Unraid:
  3. I read in. the net that it would be possible to run other OSes like Unraid on Asustore NAS Hardware. Those are basically PCs and you can get into the BIOS. Did somebody here try this out? They have a 12 NVME SSDs only / 10GB NAS in the program and I am very tempted to make an Unraid server out of this thing. https://www.asustor.com/de/product?p_id=80
  4. I know this post. And your contributions to it are appreciated. But this is basically what I was describing. There are several solutions in this post and you have to invest hours to figure out what works and what does not work. Many users are not software engineers or unix experts. We need a definitive solution by Limetech how to make this work. And this should be sorted by macos version and Unraid version since there are all kinds of problems caused by different software versions. There are also tweaks on the Mac side that should be described in one place - documentation or sticky post. And to my knowledge the main issue with this has not been addressed so far - the most time consuming part when copying files from Mac to Unraid is the moment when a file has basically been copied the the server and then some kind of halt / pause is happening - sometimes for over a minute - until the next file is copied to Unraid. Copying hundreds of small files becomes unbearable by this behavior.
  5. I don't know how many postings in the forum and on different websites I read to figure out how to get some acceptable performance when using macos with unraid. Still accessing folders on an unraid server with a mac needs real patience, the write (and read) performance simply sucks - sorry. Sometimes I "peak" to 60 mb/sec but most of the time I am around 25 mb/sec or "0"mb/sec when writing a file halts for some time (for whatever reasons). Also we have some degrading effect over time: A new booted Mac gets slower over time with copying files to unraid. With my PC on the other hand everything runs smooth and fast. And on my Synology box I also have no samba issues with the mac. I tried out all kinds of suggestions here in the forum. Some made things better, some worse, some didn't work at all. This problem is meanwhile lasting for years - and as much as love the helpful community here this is not a problem the community has to figure out. What is really needed is a "sticky" post here straight from Limetech with a definite "how to" for samba settings for mac use. And if samba is not useable for the upcoming future then give us a workaound with NFS and a "how to" for those shares (I have tried this but performance seems to be even worse).
  6. Der Durchsatz ist im normalen Array fast egal es sei denn Du hast mehr als 1 gigabit Ethernet oder wills 'mal mit ZFS herumspielen. Du kannst Unraid ja so konfigurieren, dass die HDs "schlafen" gehen, dann ist der Stromverbrauch erträglich. Was die Temperaturen angeht, so ist meiner Erfahrung nach das Hauptproblem, dass bei vielen HDs sich die Wärme irgendwann im Gehäuse zu stauen beginnt. Und dann kommt man recht schnell über 50 Grad und wenn das dauerhaft der Fall ist, dann gehen die Platten schneller kaputt. Wenn Du beispielsweise regelmässig die Parity überprüfst und alle HDs gleichzeitig über einen langen Zeitraum am Start sind, dann wird es zügig sehr warm / heiss.
  7. Nach 10 Jahren solltest Du wirklich tauschen. Seagate Platten sind mir immer einmal kaputt gegangen und habe recht hohe Temperaturen im Betrieb. Ich kann die Toshiba Enterprise Capacity Festplatten empfehlen. Von denen habe ich inzwischen 16 Stück am Start und die laufen wunderbar und deutlich kühler. 16 TB bekommst Du für etwas über 220 EUR. Wenn Du bei einem ordentlichen Händler (z.B. Alternate) kaufst, hast Du auch keine Probleme mit der Garantie.
  8. Blöde Frage: Wie macht man das? Ich verzweifle an der unterirdischen Samba Performance mit meinen Macs und blicke auch bei den endlosen Diskussionen mit Verbesserungsvorschlägen der Samba Konfiguration nicht mehr wirklich durch, zumal die meist offensichtlich nicht funktionieren. Kann ich dann trotzdem noch mit einem PC auf umgestellte NFS Shares zugreifen?
  9. I have a Gigabyte X99 Board with a Xeon 2699 v3 and 128 GB Ram For testing purposes I installed WIndows 10 on a 500 GB SSD Sata drive from Samsung. I also installed UNRAID (6.11.5) and with a 2 TB NVME Samsung SSD (installed in a 2x Port = 800 MB/S max.) and installed a VM with Windows 10 with 64 GB RAM - all 18 (36) cores are used as with the bare metal install. Geekbench 3 perfomance: Bare Metal: Single Core: 3.487 Multi Core: 45.575 VM: Single Core: 2.973 Multi Core: 39.148 In total I am losing around 15% CPU performance by using a VM. Surprisingly there is barely a difference in passing the VM trough to the CPU or using the CPU emulation. Is this in general the amount of performance loss that goes with using a VM in UNRAID? For many applications the loss may be unimportant. I wanted to build a audio workstation as VM to be able to update my hardware easier in the future. For this purpose 15% is a bit high. Should be similar for a gaming machine. Or is there hardware that is better suited for VMs?
  10. With an enterprise drive this is no problem. Endurance is much higher. Endurance for the Samsung drives is 1,3 DWDP for 3 years. Meantime Between Failure (MTBF) is 2.000.000 hours.
  11. As far as I understand it server ssds do use trim on their own and do not need any (trim) support by the os - in this case Unraid. So does this lead to any potential problems?
  12. Primarily for VM use I want to build a very fast unraid array with ssds only. The SSDs are server grade NVME SSDs from Samsung - PM9A3 Series. As I read in the UNRAID forum there are concerns regarding "wear" of SSDs. I read this also regarding PROXMOX and there was always the solution to NOT use consumer SSDs but server / enterprise SSDs that can stand extensive wear without problems. Is this the case with UNRAID, too?
  13. I had a 6 hd drives attached to 2 faulty 4 to 4 Sata Cables. OS was Windows. 5 8TB drives and one 10tb drive. DiskWarrior showed that there were connection errors to 4 of the disks which I initially attributed to failing disks and not to faulty cables. So I performed a backup of my data. I now installed UNRAID, exchanged all sata cables to new ones - as I meanwhile read that kind of error could be a connection issue - and performed Preclear (2 cycles). All Disks went with success through Preclear. The UDMA_CRC_Error_Count stayed the same as it was initially (which is as it should be) and did not get higher as it already was. To me this basically proved my faulty cables theory. However I don't know how to deal with the drives showing UDMA_CRC_Errors. Should I throw those away and get new ones?
  14. Thanks for all your help. Indeed I missed to install unassigned devices app first. Now all ist good.
  15. I can't install unassigned devices preclear on 6.10.3. I get the message "Requirements not met - The requirement of this upgrade have not been met on your system". It's am new install on a B150 mobo with an i7 7700k and 4*4TB Nas drives. Any ideas? Installing a new pool without preclear does not make sense.
  16. Correct. But Unraid is not an open source product. You have to pay for it. I wouldn't mind if the company would state "our product is ready, but driver x is not working because of the actual Linux state". Fine. Then I could decide to not install if I need this exact driver or feature. My point is that we as users are now figuring out what is not working. This is not our job. And you are right - not too many want to test beta software. Well, they are not paid for it. It's not their job. Testing by the community can only be an additional and small part of quality assurance. Well, I don't want to argue here - as I said I like the product and I will like you upgrade a few month from now.
  17. Yes, but that does not affect the need to iron out the quirks anyway. Never mind. It's not my intention to be a troll. I like the product. But it's a paid product so stuff should work. Nobody would have cared for another three month of testing to prevent those thing from happening. And frankly I can't understand that anyone came across the time machine problem at Limetech - for months. The rolling release approach that seems to be nearly as common as "subscriptions" in the industry leads basically to rolling betas that are named "releases". Everybody who has to use for example tools like Adobe CC knows what I am talking about. This is frustrating and requires far too much of my time.
  18. After flipping through this 9 pages I decided not to upgrade. Again you need to be a Linux / Unraid guru to fix (basic) stuff. I am not a noob at all but Limetech may overestimate how much time people are willing to invest to go down the rabbit hole. After months of testing and 10 RCs network and stuff like time machine backups should work without a glitch. They don't as stated here. Probably the best way to deal with this is to wait for a couple of weeks for 6.10.x , set up a complete new server and move all data over from the old one. With a 128tb server this will take forever even with 10gb. I am really looking forward to this...
  19. Seems the whole controller is not detected. Controller Bios is not showing up in boot process. I attached the diagnostics. diagnostics-20220129-1035.zip
  20. Boots again with new usb drive but LSI HBA and all drives are gone... Sigh... Probably the HBA was fried... No luck with (re)assigning drives.
  21. I shut down my unraid server and after switching it on again it did not boot. After trying all kind of things regarding to several forum posts I plugged the USB drive into my Windows machine to save the latest config. Problem: Windows can't open / access the stick and in MacOS the stick does not show up either. While I can create a new stick I only have a config backup from some time ago and after that I added another drive to the pool and deleted to old iscsi apps. I am not sure how to go on. Some advice would be great.
  22. I have problems accessing my UNRAID directories with my Mac. As soon as there are more than a handful of files in the UNRAID directory / share it's not possible to read out what's in the directory. I can see the folder / shares on the top level of the directory tree but not what's inside. I get the spinning macOS wheel. The number of files in the directory is around 5000. Nevertheless I can drop files in the directories - and by that copy stuff into the directories (slow as hell). So the main problem seems to be reading the directory with its files via macOS. All this is not happening when I access the directories from my Windows 10 Pro machine. I am pretty sure this is - again - a macOS SMB problem but if anybody can share a workaround or a solution I would be very grateful. PS: Mac runs on Catalina 10.15.7.
  23. I may have some misunderstanding here: I ran the preclear script two times on a 16TB hard drive what took way more than 4 days... After that I clicked format with this unassigned hard drive and chose XFS. Then I stopped the array and added the drive - now 4 drives and one parity drive - and started the array. Now UNRAID preclears the added drive - again. This is the third time... 15 hours to go... Isn't preclearing in advance the way to avoid exactly this procedure because the drive is "marked" as precleared? Do I miss here something or am I doning something here the wrong and stupid way? Can I stop the the third preclearing process. I don't think there is anything happening here with a third procedure...