fredhoon

After doing a bit more digging, the current version of the mlx_core module has dropped support for ConnectX-2 cards, I'm not sure what version unRAID runs however I would assume it to be fairly recent due to the kernel version. It would be a very steep learning curve for me to be able to compile and install a previous version of the OFED driver and possibly even require an older kernel / version of unRAID. I tried a ConnectX-3 card which was detected by unRAID but only provided a single network interface despite being a dual port VPI card. Both ports were in VPI / auto detect mode and despite being directly connected via a QSFP DAC cable to an identical CX-3 card in another machine, the card didn't show a link light. Compiling / resolving dependencies / installing the mft utilities in unRAID (or even mint) was beyond me so I threw the card in a windows machine and changed Port 1 from "Auto / VPI" to "Eth" & did the same for the card at the other end - now I have a 40Gb/s ethernet link from UnRAID to my Win7 Games / backup box! My cache SSDs should be good for 2-300MB/s and Turbo writes for 150MB/s so I can use at least some of the CX-3 bandwidh.

Sorry to resurrect an old thread, did you have any luck getting a ConnectX-2 VPI card working with unRAID @wheelbase or @Leondre ? From my limited linux experience it appears that the mlx4_core module is loaded and used b mlx4_en, however attempting to load mlx4_ib fails iwth error: Searching suggests that I need to compile a new MLNX_OFED kernel module that includes mlx4_ib, which looks like a steep learning curve. I'll try changing the port mode in the first instance as the Mellanox documented instructions are pretty straight forward...

Is it normal for total writes to be 10x reads when undertaking the parity copy process? I'm undertaking a parity drive shuffle / parity copy and noticed that the new parity disk writes are grossly disproportionate to the old party disk writes (~10x). I'm going from a 750G Parity disk (with 500G data / max array disk) to a 4TB disk and thus would have expected to see somewhere around 5.5-8x writes compared to reads. Read count for the old parity disk have stopped incrementing (prior to 35% copy complete), however the write count for the new drive keeps increasing (refer attached screenshots). Is this simply a case of a hiccup on "Main" dashboard or is there some reason a Parity Copy writes 10x the amount of data that it reads? tower-diagnostics-20180331-0811.zip Parity Copy - 35% Complete Parity Copy - 73% Complete Parity Copy - 91% Complete

Quick question, is there an option to auto-confirm / auto answer the "Yes" prompt in JoeL's 1.15 preclear script, or will it need to be edited to comment out that section? For reference I'm trying to string together the following 4pass drive preparation and would like to launch the script in an unattended manor... preclear badblocks (destructive) 0xaa badblocks (destructive) 0x55 preclear -W Manual confirmation of the 1st pass preclear is sensible, however I'd like the final pass of preclear to kick off automatically rather than waiting for user input. Note my *nix skills are non-existent and so far my googling has only found results for auto-answering a command, not a script.

Belated update, had to wait for all drives to finish a preclear on 2017.09.26 before I could update. 2017.09.27 is also working well at my end, and I like the added total time & zeroing time details after completion in both preclear and unassigned drives pages. [edit - turns out this was the output of the JoeL script] I've also found that the high CPU usage and unresponsive Dynamix with multiple simultaneous preclears will now complete the zeroing phase and Dynamix will recover. Previously I could only manage 4x writes and 2x reads simultaneously on my old 4core hardware, now 8x simultaneous drives zeroing could successfully complete! Thanks again for the fast and continued support gfjardim!

Cheers, I noticed that as well in the install log. I'll try and start diskinfo manually when I get home and/or reinstall with diskinfo started. Thanks for the tip! [edit] After sitting idle all day the preclear page & unassigned drives were functioning normally when I arrived home from work. I didn't restart and test manually starting diskinfo, just kicked off a preclear on 8x drives. Is is possible that diskinfo is a delayed start after a reboot and that is causing the preclear symptoms reported above?

Cheers for the quick response and 2017.09.26 update. Upgrading in place initially fixed the issues up... all drives were showing on preclear page disk details were displayed correctly in the options window however after rebooting it was back to the 2017.09.25 symptoms... no drives on pre-clear page if launched from the unassigned disks link the options window disk details were all showing "undefined" start button would not launch the preclear script and removed preclear icon from assigned disk list on main page I've since tried a clean install (remove plugin / reboot / install plugin / reboot) and am still presented with the same symptoms. Screenshots and preclear log attacehd. RACK-preclear.disk-20170927-1623.zip Before Starting preclear script from plugin: After starting preclear from plugin....

I'm getting this as well, a previous version used to show this info without issue. I'm also loosing all the pre-clear icons from Main tab after a pre-clear on a disk is started. Also the Pre-clear plugin page is now empty with message "No unassigned disks available", however all my drives are unassigned.

I've just tried the drives on a SAS3008 controller (M1215 flashed to 9300-8i P14 IT-mode) and the drives wouldn't show either. I had originally thought it was a SATA150 limitation but discounted it as the WD Raptor was working at this speed, with this in mind I checked over the drives again and the "force SATA150/300" jumper was in place at 150 for all that't weren't detecting on the 3x HBAs. Jumper removed = drives detected, no issues with P20.00.07 FW, only user error! I could have saved a heap of time by being a bit more thorough and less hasty jumping to conclusions.

The D2607 is an A11 variant, however I did try the A21 SBR (before flashing the H310) just in case, it disabled one channel as expected so I flashed A11 back. Both cards are showing the correct FW and BIOS revisions on boot and in the Ctrl+C config utility so I presume they are flashed correctly. This morning I've tried the card in another motherboard (S5000 chipset) which resulted in the same 2x drives detected and then started testing drives one by one. After running through a heap of old drives I've confirmed that both the D2607 & H310 with 9211-8i 20.00.07.00 IT mode FW will detect 8 of 8 drives connected, it's just strange that the drives they won't detect were previously detected by the OEM IR mode FW and work fine in other PC's. It appears the certain drives just won't detect with the 20.00.07.00 FW, although they work fine on AHCI motherbord SATA ports and I copied data off them in my Z170 Chipset windows PC. For info, none of my 320G or 200G Seagate drives could be detected and 2of3 1TB Seagate drives wouldn't detect on the SAS2008 with 20.00.07.00 IT mode FW (I'm going to test these on my SAS3008 card next). 2of2 1TB Samsung 7.2k drives, 3of3 1.5TB Samsung 5.4k drives, 1.5TB Seagate 7.2k drive & 150G WD 10k Raptor were all detected OK. Thanks for the help guys, time to tear down the UnRAID box and upgrade the old SAS1068/1078 controllers. And also thanks a heap to everyone in this thread for all the hard work putting the flashing guides together!

I was afraid of that after my googling came up short. I'm checking drive presence in the Card Bios (Ctrl+C config utility), GParted Live CD and left it to boot windows 7 on the desktop I was using for the UEFI shell flash (device manager and HDD Sentinel could only see the two drives). I haven't tried fdisk -I, will give that a bash on a live CD/USB, I haven't tried the card in my UnRAID box yet as I didn't want to tear it down until I confirmed at least one of the cards were working. I suppose I' ll also try querying the card with MegaCLI (or whatever the equivalent is for IT firmware). I've tried a different cable but haven't tried drives one by one as yet, I'll give that a go as well. Cheers for the pointers.

Has anyone experienced issues with P20.00.07.00 firmware on SAS2008 cards not detecting drives?I've (un)successfully crossflashed a Fujitsu D2067 A11 and Dell H310 to 9211-8i P20-IT via both sas2flsh.exe and sas2flash.efi and both cards are experiencing the same issue: Pre-flash the OEM Bios can see 6of6 SATA drives connected across both channels via fan-out cables. Post-flash both cards can only see the same 2of6 drives across both channels regardless of fan out cable connected port & channel. If I change the port drive connections around the cards will still only see the same 2x drives. Test drives are 2x Seagate 1TB, 1x Samsung 1TB & 3x Seagate 300GB, after P20 crossflash only 1x Seagate 1TB & 1x Samsung 1TB can been seen by the controller Bios At first I thought it might be an issue with the modified D2067 A11/A21 firmware however the exact same issue is present on the H310 card, same drives. I'm going to try again with P19 or P18 firmware and without a Bios, maybe even try a fresh download of the 9210-8i P20 firmware from the Broadcom website. It's a strange issue and I can't logically explain / troubleshoot the symptoms, has anyone experienced this before or able to offer some pointers to help me troubleshoot further?