Last_phil Posted September 9, 2019 Posted September 9, 2019 1 hour ago, martinayotte said: It is not "fsk" but "fsck" ... 🤦♂️🤦♂️🤦♂️🤦♂️🤦♂️🤦♂️🤷♂️🤷♂️ Thanks a lot it help to type the right command everything ia back om track.... Just need to remember my password now. 👍
gprovost Posted September 10, 2019 Author Posted September 10, 2019 17 hours ago, dazhan98 said: I own a helios4 batch 3 and just curious, does anyone know how long it usually takes for an armbian release of a debian update? they just released 10.1, I was thinking of waiting for a Helios4 image before setting everything. You don't have to wait, using apt-get upgrade will update your system to the same packages version (or even more recent) than the latest Debian Release image. See Debian Release update as just a new image built with the latest packages available.
dazhan98 Posted September 10, 2019 Posted September 10, 2019 9 hours ago, gprovost said: You don't have to wait, using apt-get upgrade will update your system to the same packages version (or even more recent) than the latest Debian Release image. See Debian Release update as just a new image built with the latest packages available. Oh legend! Thanks, I'll start setting it up tomorrow night then. Been looking forward to it for a while now.
uiop Posted September 10, 2019 Posted September 10, 2019 On 9/2/2019 at 5:22 AM, gprovost said: Regarding the advertised max raw capacity of 48TB, this doesn't take into account Port Multiplier usage. It was just the max raw capacity using 4x of the biggest NAS HDD available at that time on the market. Does this mean I can use 4 x 16TB? The wiki is a bit confusing, it says: "Max Raw Capacity 48 TB (12 TB drive x 4)" and "Max Single Volume Size 16 TB". What about 2x12 and 2x14 for example?
Igor Posted September 11, 2019 Posted September 11, 2019 7 hours ago, uiop said: What about 2x12 and 2x14 for example? 16TB is partition/volume limit. Whatever goes below that is fine. You can mount drives separate, but can't join them into one volume that is larger than 16TB. It's 32bit OS/hardware limit. 1
gprovost Posted September 11, 2019 Author Posted September 11, 2019 @uiop The 32-bit architecture limitation is actually on Linux page cache which is used by file systems, therefore limiting the max usable size of a partition or logical volume to 16TB. This doesn't stop you to have several partition (or logical volume) of <16TB. Here an example taking in consideration 4x 12TB HDD : If you use mdadm and you setup a RAID6 or RAID10 array, you will have an array of 24TB of usable space . You can then create 2x partition of 12TB or any other combination that would max out the array size, till a each partition doesn't exceed 16TB. If you use lvm and you setup a Volume Group (VG) with the 4 drives (PV), you will get a volume group of 48TB of usable space. You can then create as many (max: 255) Logical Volume (LV) to max out the VG size till each LV doesn't exceed 16TB. Actually a good approach is to use LVM on top of MDADM Raid, this way it gives flexibility to resize your partition, or i should say Logical Volume (LV). After you can of course neither use mdadm or lvm, and just setup individually each disk... till you follow the rules that you can't create a single partition bigger than 16TB, but you can create several partition. Hope it clarifies. 1
Doni Posted September 11, 2019 Posted September 11, 2019 10 hours ago, gprovost said: If you use mdadm and you setup a RAID6 or RAID10 array, you will have an array of 24TB of usable space . You can then create 2x partition of 12TB or any other combination that would max out the array size, till a each partition doesn't exceed 16TB. @gprovost Do the tools to create filesystems like mkfs.ext4, mkfs.btrfs and so on, prevent the user to create partitions bigger than 16TB knowing that this is a 32-bit architecture? Or is it up to the user to know this limitation? Also couldn't btrfs be used to beat the 16TB limit on 32-bit architecture?
gprovost Posted September 12, 2019 Author Posted September 12, 2019 10 hours ago, Doni said: Do the tools to create filesystems like mkfs.ext4, mkfs.btrfs and so on, prevent the user to create partitions bigger than 16TB knowing that this is a 32-bit architecture? Or is it up to the user to know this limitation? That's a good question and I don't have a details list of partitioning & fs tool behavior. Some tool will prevent, but other will let user create partition bigger than 16TB and the issue will only show up later (e.g during the fs inode table initialization or anything trying to access a block that goes beyond the 16TB region). I must admit that it would be a very useful investigation to test all the use case with the different tool available. As for btrfs, apparently on 32-bit architecture it is recommended to not create volume bigger than 8TB because some brtfs tools might not work properly on 32-bit arch with bigger partition. I'm just reporting what I read online, I haven't tested.
devman Posted September 13, 2019 Posted September 13, 2019 (edited) 21 hours ago, gprovost said: As for btrfs, apparently on 32-bit architecture it is recommended to not create volume bigger than 8TB because some brtfs tools might not work properly on 32-bit arch with bigger partition. I'm just reporting what I read online, I haven't tested. Oh <edit: expletive actually deleted>. umm, do you have a source for that? Edited September 13, 2019 by devman
gprovost Posted September 13, 2019 Author Posted September 13, 2019 @devman https://btrfs.wiki.kernel.org/index.php/Gotchas#8TiB_limit_on_32-bit_systems
uiop Posted September 16, 2019 Posted September 16, 2019 On 9/12/2019 at 8:24 AM, gprovost said: As for btrfs, apparently on 32-bit architecture it is recommended to not create volume bigger than 8TB because some brtfs tools might not work properly on 32-bit arch with bigger partition. I'm just reporting what I read online, I haven't tested. That's a pity, but good to know. So if I want my 16TB Volume to have checksums of file contents (against bit-rot), there's only ZFS left, with its own problems (not in kernel, eats a lot of Ram).
devman Posted September 17, 2019 Posted September 17, 2019 21 hours ago, uiop said: That's a pity, but good to know. So if I want my 16TB Volume to have checksums of file contents (against bit-rot), there's only ZFS left, with its own problems (not in kernel, eats a lot of Ram). And I need to find a way to migrate my current install. I turned 3x 8TB into a 16TB raid-5 BTRFS array, and am somewhere around 8.5-9TB in use. I thought I had a lot of headroom, but if the tools won't work...
jimandroidpc Posted September 18, 2019 Posted September 18, 2019 Is the CPU cache also ECC ? Does that matter either way in this device?Sent from my LM-G710 using Tapatalk
gprovost Posted September 18, 2019 Author Posted September 18, 2019 @uiop @devman Well it would be great to cross check first what I reported about BTRFS tools on 32-bit arch. Maybe there have been some improvements done by the BTRFS community. @jimandroidpc Yes L1 and L2 cache is parity protected on Marvell Armada 38x. Well it wouldn't make sense to have ECC RAM with a CPU (SoC) that doesn't have at least cache protected with parity check mechanism.
Mangix Posted September 19, 2019 Posted September 19, 2019 Anyone able to get hardware acceleration working with OpenSSL? I installed cryptsetup and ran benchmark. I saw that the crypto engine was being used as it generated interrupts (/proc/interrupts) but not with OpenSSL. I recall cryptsetup giving me 80MB/s on some AES setting and 20MB/s when I did rmmod on marvell_cesa. I can't get it working with OpenSSL though.
Mangix Posted September 19, 2019 Posted September 19, 2019 Bah kernel issue. lsmod says only af_alg is loaded. algif_hash and algif_skcipher are also needed. Actually I'm not sure about the former since hashes are a bad idea to hardware accelerate.
aprayoga Posted September 20, 2019 Posted September 20, 2019 @Mangix have you check the wiki, https://wiki.kobol.io/cesa/ ? There are plenty of informations over there, for example the benchmark result and what kind of cipher can be hardware accelerated.
Mangix Posted September 20, 2019 Posted September 20, 2019 Those instructions won't help when the necessary kernel modules are missing.
gprovost Posted September 20, 2019 Author Posted September 20, 2019 @Mangix algif_hash and algif_skcipher are already compiled as module, so there aren't missing : Kernel 4.19 https://github.com/armbian/build/blob/master/config/kernel/linux-mvebu-next.config#L5113 Kernel 4.14 https://github.com/armbian/build/blob/master/config/kernel/linux-mvebu-default.config#L5161 Did you configured OpenSSL to offload on AF_ALG ? https://wiki.kobol.io/cesa/#configure-openssl
Mangix Posted September 20, 2019 Posted September 20, 2019 Then why don't those modules appear on my Helios with lsmod? Running Buster. Only af_alg shows up.
toe Posted September 21, 2019 Posted September 21, 2019 root@helios4:~# sudo picocom -b 115200 /dev/ttyUSB I am new to linux and tried to enter this command into the terminal but i got back the following issue below: FATAL: cannot open /dev/ttyUSB: No such file or directory How do i go about trying to connect to the dev/ttyUSB?
gprovost Posted September 21, 2019 Author Posted September 21, 2019 Most probably not loaded because no software is calling the AF_ALG API. Again, did you configured OpenSSL to use AF_ALG ?
NicoD Posted September 21, 2019 Posted September 21, 2019 11 hours ago, toe said: root@helios4:~# sudo picocom -b 115200 /dev/ttyUSB should this not be : sudo picocom -b 115200 /dev/ttyUSB0
lanefu Posted September 28, 2019 Posted September 28, 2019 Heliso4 Gen2--No longer powering up drives I came to my office this morning and i noticed my helios4 was running fans at full speed and not responding. I power cycled.. fans ran in burst intervals.. serial console stopped here: Spoiler igh speed PHY - Version: 2.0 Detected Device ID 6828 board SerDes lanes topology details: | Lane # | Speed | Type | -------------------------------- | 0 | 6 | SATA0 | | 1 | 5 | USB3 HOST0 | | 2 | 6 | SATA1 | | 3 | 6 | SATA3 | | 4 | 6 | SATA2 | | 5 | 5 | USB3 HOST1 | -------------------------------- I unplugged power to all hardrives. Helios4 booted up. I tested each drive individually on another machine and they are all fine I plugged in 1 harddrive.. Helios4 booted, but did not see drive... I heard a bit of electrical squeeling from somewhere (couldn't identify source) I tested both pin pairs on the power supply.. both read 12.3 volts Any advise on how to proceed from here?
Heisath Posted September 29, 2019 Posted September 29, 2019 Can you / have you measured the 12V rail when the drives are plugged in? This sounds like a faulty supply or on board regulator which fails under load. The squeeling could be a switching regulator desperately trying to work...
lanefu Posted September 29, 2019 Posted September 29, 2019 6 minutes ago, count-doku said: Can you / have you measured the 12V rail when the drives are plugged in? This sounds like a faulty supply or on board regulator which fails under load. The squeeling could be a switching regulator desperately trying to work... Not yet, i was trying to hold off on full disassembly I'll do that next...
lanefu Posted September 30, 2019 Posted September 30, 2019 7 hours ago, lanefu said: Not yet, i was trying to hold off on full disassembly I'll do that next... Okay takin it apart is actually easy ha... So i put my test lead inside the 12v connections into the back of the 4 pin molex connectors on the board. At powerup ..even with 1 drive, voltage is dropping to 9 or 10 volts. and the fans sometimes stutter. It goes back up to 12v after the drives fail to spin up 5v pins on molex seemed fine I think the noise is coming from the board, but its not a loud as earlier. Also I tried with different drives on both ports and got the same behavior
lanefu Posted September 30, 2019 Posted September 30, 2019 FYI I just powered all the drives with a separate power supply and they're all online
gprovost Posted September 30, 2019 Author Posted September 30, 2019 @lanefu It clearly looks like a dying PSU. Expected measured value, on 5V rail: 4.90 V - 5.20 V on 12V rail: 11.90 V - 12.5 V If 5V is outside that range, that mean the onboard regulator is failed (board problem). If 12V is outside that range, that mean the power supply is failed. So you confirm the issue on the 12V rail ? If yes then we will send you a new PSU.
Recommended Posts