starcat

Ich würde im Server-(Haupt-)Gehäuse die interne backplane über einen LSI 9305-16i anschliessen (über drei SFF-8643 zu SFF-8087 Kabel), die anderen zwei externen JBOD Gehäusen über eine LSI 9305-16e anschliessen (jeweils dual link über zwei SFF-8644 auf SFF-8087 Kabel je Expansion JBOD Gehäuse). Beide Karten sind PCI v3 8x und hätten genug power. Nimm PCIe v3 und SAS-3 Controller damit du intern und extern mehr speed und Bandbreite hast. In den JBOD-Gehäusen könntest du jeweils einen slot blech Adapter verbauten (extern SFF-8644 auf SFF-8643) und dann intern wieder SFF-8643 auf SFF-8087 zum backplane gehen (dual link). Von Supermicro gibt es dafür direkt ein slotblech Adapter von SFF-8644 (extern) auf zwei SFF-8087 Kabel, die dann direkt auf die backplane gehen. Dann nimmst du insgesamt 4 SFF-8644 auf SFF-8644 Kabel, 0.5m lang. Wäre die sauberste und professionellste Lösung.

For instance these

Selling two of these, new and unused Require a PCI v4 mainboard with a PCIe 4x and an adapter card like the Delock 90081 or U.2 Not compatible with PCI v3 slots. Huawei ES3500P V6 NVMe SSD U.2 Festplattenkapazität - 7.68TB Produkttyp - NVMe SSD - Solid-State-Drive NVMe High-Endurance Hersteller - Huawei Herstellerbezeichnung (SSD) - Huawei ES3500P V6 NVMe SSD Technologie - SSD NAND Flash - 3D TLC Format - U.2 PCI Gen 4 Seq Read (at 128KB) - max 7.2 GB/s Seq Write (at 128KB) - max 4.2 GB/s R/W IOPS at 4KB - max 1.600.000, max 230.000 Asking price EUR 900 for the piece

Exactly, wont catch up. Got replaced by NVMe over U.2 and U.3

Anyone with some WD120EDAZ drives wanting to part with, please let me know. Preferable in the EU. Thanks a lot.

yes, or run the command from a VM on another host that stays open.

dd bs=1M if=/dev/zero of=/dev/mdXp1 status=progressI have 6.12.10, so the above should be the right command to zero the disk, right? X is the disk number, ofc. Do I need to run dd in Maintenance Mode or not really?

Asking basically how exactly to clear (zero) the disk, before I configure it out of the Array? Everything else is clear to me.

Yes, I will do it manually. Did it lots of times in the past. Forgot whether the disks need to be zeroed (dd) or something else, before I re-create the Array without the disk.

There was a way to remove a disk from the Array (new Config and leave out the disk-to-be-removed) without parity recalculation. This is needed in order not to be exposed without parity (and protection) during a rebuild. How should I prepare the disk to be taken out? Anyone remembers the procedure?

Have now three PCIv3 LSI 9300-8i in IT-mode in the PCIe 8x slots and everything is running fine. Didn't use the onboard SATA ports as the SAS ports are hot pluggable as well as 12Gbps on the LSI 9300-8i.

Thanks a lot, going for it. Will mate it with an Intel Xeon E5-2680 v4

Hi there, any experience with the Supermicro X10SRI-F board running unRAID? It is an LGA2011-3 Single Socket Intel C612 ATX DDR4 1TB Mainboard Cheers starcat

Researched the AOC-SASLP and SAS2LP and it looks like they do not work reliable, or at all... Looks like the LSI 9211-8i (PCIe x8) is still the best option today? Firmware P20 and running in IT-mode.

Hi guys, running an old Supermicro X7SBE mainboard with two Supermicro AOC-SAT2-MV8 PCI-X boards for 2x 8x SATA devices. I would need to connect one more HBA for another 8x disks and as the mainboard has one unused PCI-E 4x and one PCI-E 8x, I was thinking of an additional LSI 9211-8i (IBM M1015, flashed) HBA or maybe a Supermicro AOC-SASLP-MV8 (Marvell 6480 host controller) Supermicro AOC-SAS2LP-MV8 (Marvell 9480 host controller) What would you recommend as the best option today for this setup? Running unRAID 6.12.10. Thanks a lot for any hints.

Thanks a lot for your hints, helped a lot!

I need to be able to use root user from another linux box to rsync files from the other linux box to the unraid server. For this I need the NFS disk exports to have the no_root_squash option in them. I can modify /etc/exports and execute exportfs -ra and this solves my problem of being unable to chown from within rsync as the NFS export has all_squash in it.

Hello All I was looking for the no_root_squash option for the NFS exports but could not find it in the GUI. Before I copy over the /etc/exports file after each reboot, is there any other way to specify it for the NFS exports? Thanks a lot!

Anything I can check my data disks that are all XFS formatted? Thanks again.

I am sure that at least one unclean shutdown happened. Also there were never ever sync errors before. This, I am certain. I just can not say for certain, that the sync errors were caused by the unclean shutdown... You are probably right though. I don't have checksums, I am now thinking about that. But, I am using XFS on all of the disks - any ideas as to what and how I can check?

But I am not 100% sure. Is there a way to have a deeper check to be certain before a kick the parity check with correction?

Hi guys, my server is running without issues in a rack in a Norco 20 disk chassis. I suddenly had an issue with the UPS and the server was unproperly shutdown after a power failure. Later, during a Parity-Check, I noticed 1295 sync errors, which stay constant and don't move (if I restart the Parity-Check, without correction´). I did not correct the parity so far. Now, what would be the best way to find out, if I just Parity-Check and correct the parity or have to deal with errors on some drive? I never had any hardware issues and it doesn't look like a cable, etc. issue. The server doesn't move at all. I have 17 devices (ST8000DM004) in the array and two WDC_WS121KRYZ as parity drives. I have 6.12.10 installed.

Thanks a lot guys!

Ok, I see, so probably my preclear_disk.sh v1.7 isn't writing the proper signature and the disk do not get recognized as pre-cleared in the newer array. Which version or other tool do I need to write a signature to the disk (it is already pre-cleared with the old script). Btw, what happened to Joe L., must have been at least 10 years..

No, not all, not doing anything. Just pre-clear and inserting the pre-cleared disk into the production array. Thanks for confirmation, was thinking I am missing something.