Wody

I don't think they sell 4-pin without cable because of the weight, the connector would bend and could break. Best solution would be to use the 4-pin with cable, and then use double-sided tape or something to attach it to the side of the case somewhere.

That means the PCIe slot is physically big enough for an x4 card (with 4 lanes), so an x8 or x16 card that have a bigger connector do not fit, but an x1 or x2 or x4 card will fit. However, only 2 lanes (x2) are connected to the system, so the other 2 lanes have no connection. So that means a card with an x1 or x2 connector will work at max capability, but an x4 card won't be able to use 2 lanes, so it only will have half the bandwidth as maximum.

Most consumer boards and processors don't support bifurcation (splitting of PCIe lanes), so you need the bifurcation to happen on the expansion card itself with something called a PCIe switch. Try googling 'm.2 to pcie switch' and it should come up with some. The ones i found used the ASM2824 chipset and use PCIe 3 instead of 4 but that should get you started.

The manual says 'sas cable' so that isn't very useful, googling 'Dell PV MD1400 cable' (without the quotes) brings up cables like the Dell GYK61 which is a 12Gb HD-Mini sas cable, in other words, an SFF8644 cable. Wikipedia has a picture of it. Since the R730 can have multiple different controller cards, from your post it is not easy to guess which it is, however, you have multiple options. If you already have a SAS controller, you can buy cables with SFF8644 on one end, and whatever you have on the other end that you can connect directly (SFF8644, 8643, 8087 etc.). If you want to keep things neat, you'll need a card with the same SFF8644 connector, so something like a Broadcom/LSI 9300-8e or newer. As an alternative, you can also use any internal SAS controller with a SAS expander, such as the Adaptec AEC 82885T which has both external and internal ports, and they are very cheap used.

Usually Trim isn't enabled by default, which can be set up in the scheduler and in the options for the drive with autotrim. Not enabling it means the same sectors get used which causes degradation and early failure. So enabling those if they aren't will help a lot. As for the new drives, performance is different between them but doesn't really matter a lot usually, same with PCIe, its more important how long they last, but that is for all models of the 990 the same, so the advertised difference is only speed. If the docker-files themselves don't change a lot, you could also set them up, make a copy of the folder to the backup pool, but rename them so the cache-dockers are used, and if they fail, remove the cache and just rename the backup folder to the regular name.

If you really are worried about speed, and want to throw money at it, the 9500-16i is PCIe 4.0 and since each generation doubles the bandwidth it would get the same result on an x4 slot as the 9305-16i on an x8 slot, provided it is PCie 4.0 of course, which can be an issue with some x4 slots, especially if they come from the chipset.

I have a Zhenloong LF24-12G which when I bought it also had a 3x36 expander version. I chose not to buy it, because those chips need firmware, and if there is something wrong there is no access to it. There is also an SDK so if you can program you could fix things, but again, no access. I'm sure they'll work fine though, and those cases are very easy to work with. Only thing I haven't figured out yet is how to get the fans out of the carrier, but I haven't tried very hard.

You do want a cable with two connectors, so an SFF-8654 8i cable. They make a SFF-8654 4i cable too, but the pinout is different, and it is much smaller so that won't fit. However, you should only connect one of the two connectors to the expander. Many expanders do have two connections, but those are for redundancy so you can connect it to another SAS controller, but unraid doesn't support that.

The 9200 series uses sas2flash, the 9300 series uses sas3flash. They are not compatible with each other, so using the wrong one with the wrong card won't find anything as you found out. For both sas2flash and sas3flash the commandline is simply sasXflash -f firmwarefile, or for the bios sasXflash -b biosfile where X is obviously 2 or 3. Both versions have no HTML embedded and are only a single file, so whatever you're doing is not calling the right sas2flash. These days it is recommended to do it from an UEFI shell, although it shouldn't matter much.

Since it shows no free space used or free, and the many errors in the log, your USB device has failed and needs to be replaced.

You'll get the full speed on the drives and on the network, so not too slow, but the single core speed is lower, so that means if you visit the unraid page for the server, or go to another page, it won't be instant, but may take a second. That's perfectly fine, and most people wouldn't notice, but if you're used to it being instant on other computers you'd notice. Here it is supposed to work the same way, but the convertor they put in had to be replaced three times, and the power-supply failed twice too during only a year or two. So I bypassed that device and extended and connected the cable directly to my switch where its running for years without any issue. Anyway, for Intel server boards, they do support bifurcation, and the x11spi-tf has slot6 which supports x4x4x4x4 for sure, not sure how the others handle it because of the switch and reserved pcie lanes for the chipset, but the other one probably also supports it. The system block diagram in the manual shows which slot is connected to which IIO and you set the bifurcation for those in the bios.

Point is, it will work, but if you visit the dashboard or something it may not feel as fast as it could be. But a lot depends on what you're used to. All 10gbit RJ45 get hot and use a lot of power (up to 5w per port or so, so 10w per connection), fibre uses up to 1 watt or so no matter the speed (but then you also need to get the SFP+ and a network-card so it may not matter that much). Point here is, there always has to be some airflow in the area of the network-card. So if you have that, then it's fine, but its good to know in advance.

See this thread for recommendations. I use a Samsung Pro Ultimate microSD card with sd-adapter which I like a lot, its a lot faster than my USB sticks (which are older). I keep copies of the unraid-installations on the server so I don't really need to backup and haven't had a reason to replace the card yet, but it did work great while testing that.

I use the X11SPL-F which is basically the same board, except the X11SPi-TF has a better chipset for the network, and a different PCIe configuration. Only thing you have to keep in mind is that if you use x8 slot 3 then the x16 slot 4 will also work at x8. Most of the time that won't make a difference for performance though. Also slot 1 is only x4. The board does support bifurcation, for streaming you probably want to add a videocard, which can influence how you work with things, since it already has a VGA-controller on board. I've used mine with a 6138 and it worked okay but I wasn't happy with the performance, so I switched to a 6248. Yours is supposed to be network-optimized so benchmarks show it may be slower than both but it'll probably work better for what you want to do with it. Also keep in mind that 10gbit copper requires a lot of power and the chips become very hot so you have to make sure you have enough airflow over the heatsink near the slot which can be an issue if your x16 cards are big.

In the network options you can change what MAC goes with what ETH connection, so you can make one of the ETH4 and 5 become ETH0, does that help?

Correct. See the Seagate FAQ.

That means the switch is not compatible with the cable/SFP-module you're inserting into it. Try connecting the cable between two computers directly, and see if they negotiate at 10gbit. If they do, your router doesn't like the brand of the cable, or the cable itself. Also, SFP tends to stay in the same state once you insert something sometimes, so try shutting down both computer and switch, remove the cable between them, start them up and once up, insert the cable first in the switch, then in the computer, and see if it stays at 1gbit.

1. Yes, although technically you're not supposed to since it might not be a real trial. 2. The license is linked to the USB flash-drive it is currently on, so you can't copy that, but assuming you didn't change the start-menu, it is a case of renaming the config folder, then moving it to flashdrive 2, move the config folder from drive 2 to drive 1, move the licensefile from drive 2 to drive 1, and rename the config folder on drive 2 (so basically switch the config folders, but keep the key on the original flashdrive). 3. Technically Beta and RC versions could have errors that make things not behave like they should, but unraid is very good in keeping things stable, so yes.

I figure they are optimized values, so you won't see them in the real world, but they are probably all skewed in the same way, so the comparison between them should hold. I'm not sure if I could measure it reliable, I do have a watt-meter but since the difference between the 9500-16i, 9400-16i and 9300-8i that I have is only 2w each, it would fall in the error-margin.

If your drive doesn't spin up, it is probably dead, however, it could be the 3.3v issue, so make sure to use a molex to sata adapter to see if that is the issue. If it doesn't go away, 0x4 0x2 means logical unit not ready, initializing command required. In other words, the drive needs to be initialized or formatted. I don't see any initialize command, so probably sg_format should work.

I haven't looked up the 9305, but the 9400, 9500, and 9600 (and probably 9700 they are working on currently) are all HBAs, so no IT/IR anymore. The 9400 has two firmwares, one for SATA, SAS and NVMe, and one that lacks NVMe but has bigger buffers and can handle more commands at once. To flash, you'd use storcli. The 9500 had mostly one firmware, until P30, now it also has one with and without NVMe. It also uses storcli, but a different version. The 9600 only has one firmware-package with everything, it needs storcli2 to flash. Advantages of the 9400 and 9500 are that each uses less power than the previous generation, the 9600 uses more since it is a SAS4 card.

Honestly, I'm not sure that is the case. I've tried re-creating the situation, on my X10SL7-F which has the same SAS-chip (although revision 5 instead of 3) with an 8TB HUH728080AL4200 which should be the same type, and it just works. So I started reading your logs some more, found some weird stuff like SAS addresses going 44332211... which turns out to be your SATA drives, so that wasn't anything, but the 'bad' drives do report something, which is their status. To be specific, after a drive is connected, there is a report 'Sense key 0x2' which comes out of the SCSI specification to indicate something special is happening, which in this case means Not Ready. Furthermore, it reports ASC (additional sense code) 0x4 which is something with the hardware, and an ASCQ (additional sense code qualifier) of 0x1b which means Logical unit not ready, Sanitize in progress. In other words, the drives seem to be working, but you have to leave them alone for about a day or so to remove old data. If you have the unassigned devices plug in, you may be able to get the drive page, which has a self-test tab, with a self-test history. If you open that, it should say how long an extended self test takes, which should be the same as the sanitize-process it is doing. Leave it alone for that long, and see if it works then. It should report everything okay. I have no idea however, if it will start over santiizing if you change slots, controllers or anything else though, so make sure to keep that the same after a drive is complete to see if unraid recognizes it, and if it doesn't, restart the computer and see if it comes up then. In case this status doesn't mean it is already doing self-sanitize, but it is saying I'm not letting you do anything until you sanitize me, you can use (using sdd as an example): sg_readcap -l /dev/sdd which will show something about the protection of the drive and the capability (not sure what it all means though) and to sanitize the drive use sg_sanitize --type /dev/sdd where type is block, crypto, fail or overwrite. If you do sg_sanitize by itself it will explain some of it.

SAS drives are special in that even if the drive is completely dead, as long as the electronics or firmware work, they can be interacted with. But not like SATA drives, because they are not SATA. So, try something else first, see if they can report their status, and run: smartctl -d scsi --all /dev/sdd This will show all smart data for device sdd like in your screenshot. Also, you should post diagnostics.

The errors that Fix Common Problems reports are also visible in the log, so the plugin reports them, but you don't need the plugin for it. As to the errors, Machine check events are events that are reported by the processor. It puts them in a file which you can view by using the terminal or use ssh (if enabled) to the server and type something like cat /var/log/mcelog I'm not sure what would show up though, since my machines don't report any errors, and so the file is empty. If the file doesn't tell you what you need, you can use mcelog to decode it. Again, no idea what it will show. If you do want us to see it, you can dump it in the system log with mcelog --syslog and then re-run diagnostics and upload those. Also, the bug-report-forum for pre-releases is here but your issue is a hardware-issue, so unlikely to be related to the unraid version,

Since they mention firmware, it probably isn't a physical limitation, but just that they don't want to make the firmware for old unsupported devices. You might want to look into Broadcom devices too, they seem to be more widespread and cheaper. The Adaptec software is more userfriendly, but for an HBA that doesn't really matter. My backplanes have SFF-8087 connectors or SFF-8643, so a lot less cables, but otherwise its pretty much the same. Except of course this extra 8-pin plug you have, which is probably a SGPIO sideband connector to communicate with the backplane, on how and when to blink the lights. The SGPIO and IBPI are two different patterns for it. My backplanes don't have that, since the signals travel through the connectors, but SATA connectors don't have those signals, so you need a sideband-connector. Of course with SATA you can blink leds too, like the activity LED, but not with as much variation as the backplane is capable of.