Both of the cards I was referring to will work with OMV 5. I'm sure because I use 'em with OMV.
Answering about speed, I can't confirm if they will allow 200MB/s speeds as my WD 4TB drives cannot achieve it. I'm getting about 125 MB/s when copying files between internally connected WD40EFRX and another WD40EFRX connected to IOcrest M.2 E key to SATA adapter. Not sure if higher capacity REDs will get higher speed numbers, never used one of those.
I said that drives will have available throughput of ~200MB/s (per HDD) using those controllers not that those are the speeds that your WD REDs will hit. RED's are not the fastest drivers out there and are not meant to be. But if you connect 4 RED's to 4 SATA port controller sitting in PCI-e 2.0 x1 slot you won't even hit 125MB/s as speed will be limited to ~100MB/s (if all of the 4 drives are being used simultaneously).
I cannot guarantee if 10TB drive will work with JMB582 controller but I see no reason why it should't. Only drive capacity limits I've read about were referring to old Marvell controllers. Newer ones like ASmedia's and JMB's should support it.
Here is a shot from copying 15GB files in total to WD40EFRX connected to IOcrest M.2 E key controller:
If I can get an LSI SAS HBA card that is the same price per port as a no name SATA card, why would I choose the no name card?
True. So find him LSI SAS card for PCI-e x1 slot. Never saw such a thing.
What do u think about this ? https://www.amazon.de/dp/B07NT…apa_i_q11cGb9D622PQ?psc=1
It should work as it's old marvell chip and well supported under Linux. But as I said before you will limit your maximum throughput with it. If you are fine with it that's a good cheap choice. One thing to keep in mind is that those old marvell controllers tends to overheat under heavy load and small heatsinks on those boards are often not enough.
If you want to go this route (one card with 4 ports) I would wait for confirmation that new Asmedia ASM1064 controller is supported by linux kernel.
Cards with it are already there:
But haven't found anything about linux support for that controller. Should be a matter of time though.
So motherboard is an asrock j5040 whit pcie 2.0
I'm using older model from Asrock, namely J4205-ITX and also found 4 sata ports to not be enough. I can tell you what I use and why I've chosen this way of expanding number of SATA ports.
Having PCI-e 2.0 x1 limits your throughput to 500MB/s (theoretical, real speed is lower, probably below 400MB/s) so using sata controller with 4 ports and 4 HDD's connected to it would be a bottleneck even for old HDD's, not even to mention SSD's. So I've went with sata controller with 2 ports, namely Syba SY-PEX40039 (same model number for brand called IOcrest) because it uses same Asmedia (Asus' brand) ASM1061 controller as our motherboards and it's cheap. It works in my setup for almost 5 years now and I haven't had single issue, speed is same as connected directly to motherboard. That's to have 2 additional ports.
Now, our motherboards does not have any more PCI-e slots, but they have spare wifi port, specifically M.2 E key slot (which is exactly PCI-e x1 signal with different slot) and clever Chinese from IOcrest made a little thing to convert this M.2 E key to additional 2 sata ports:
available also at aliexpress directly from manufacturer: https://www.aliexpress.com/i/4000385259838.html
Bought this tiny thing about half a year ago and it also works very well in my setup. This way I'm using all available PCI-e lanes in my motherboard and each drive has about 200MB/s throughput for itself. For home usage it's enough, at least for me.
I've mentioned some specific brands and stores few times so I would like to make clear than I'm not connected with any of them in any way. Just shared my own experience with those products.
Hello I want to upgrade my storage space qnd i need a pcie 2x1 card i want to add 3 more hdd. Can u recomand one pcie expansion it is best to choose sas or sata ?
To recommend anything we need more information, especially your motherboard model or your PCI-e slot version. One thing I can say about SATA controllers is that I would vote against using more that 2 SATA ports controller in PCI-e 2.0 slot (common in mini-ITX boards) but if you have PCI-e 3.0 x1 you could easily go 4 ports SATA controller as throughput will be enough for 4 HDD's. Different story if you want to connect SSD's.
This is OMV 5.5.22-1, I just umounted (in the GUI) a volume to fsck it due to file system errors, now when I attempt to remount it (again in the GUI) it is mounted to /srv/dev-disk-by-uuid-???? instead of the previous location /srv/dev-disk-by-label-<volume label>. The contents of the file system are all intact, but all the shares will be wrong. Is there any way to enforce the old behaviour?
Not quite but if you need to do it once you can follow what I wrote here:
I'm sad about that. Thanks for the reply!
You can always downgrade to openmediavault_5.5.19-1, mount/remount your filesystems - they should mount by label, if not restart after downgrade and check again.
Downgrading could be performed as simple as:
After you're done you can always upgrade back to current version. It's 3 minutes job.
I have a possibly stupid question to ask but I really can't figure out the drive size differences myself...
I'm using OMV with few EXT4 formatted 4 TB drives, which transfers for 3,64 TiB capacity shown in "Storage->Disks" - I do see a difference between TB and TiB so that's fine,
But where does the value of 3,58 TiB available storage (displayed in "Storage->File systems") comes from?
Gparted displays correctly 3,64 TiB for each drive but df -k (not using df -h this time due to rounding to one significant figure) shows 3844640564 1-k blocks, which in turn equals to 3,58 TiB.
Reserved block count is 0 - guess it's default when creating EXT4 partition in OMV GUI.
What am I missing? Can someone explain?
I know, it's not a big deal but the fact I can't figure it out is a real pain in the a** ;d
Sorry for resurrecting old thread but I'm having same problem with nvme ssd as the OP.
Already tried installing smartmontools (v. 7.1.1), as well as kernel from debian-backports. Smart information tab shows device field as "/dev/nvme0n1" and serial model on that tab is missing (says n/a), therefore there are no information under "Attributes" tab. That's strange because smartctl gives correct data, no matter if I use "smartctl -a /dev/nvme" or "smartctl -x /dev/nvme0n1" or even "smartctl -x /dev/nvme0n1p1" - output is the same and its correct.
Any idea how to get smart info from omv webpanel for nvme ssd?
P.S. There is one more thing, is that correct? Why do those values differ?
First I would like to apologize for not keeping up with the thread for so long but I was off for few weeks.
Secondly thanks for answering @ryecoaaron!
I want to have DE because I store almost all my files in nas and when I need to do something with them, like creating/extracting archives, editing video containers etc. It runs faster when command is being executed on NAS itself instead client PC. Probably gigabit ethernet is a bottleneck here. I'm more "windows guy" so doing this in DE is much more convenient for me than typing commands in shell.
That's not something crucial for me but would be nice to have. Additionally my NAS is standing next to my desk and it's connected to the same monitor which I use daily with my PC.
And during tests I haven't run into a single problem of that connection. That's why I ask what exact issues might to occur.
I would like to install OMV on Debian 9 with DE (LXDE for example) and threads I've found about that type of installation are full of answers that this is not recommended, that it leads to problems.
Can someone explain it to me what is the problem with OMV alongside DE? What can happed if I try it? I know that OMV is meant to be used headless but I made clean install just for testing purposes and it was working as intended or I wasn't using it long enough to encounter any problems. Is it lxde-session (in this scenario) which causes problems or what?
And what if I would set grub to boot in tty and run DE through startx command only if I need it?
Would be greatful if someone could shed some light on this matter for me.
Thank you for quick replies.
It seems that's 2:1 to leave it alone and not make any changes to weakref.py. Searching the forum with this file name, as @macom suggested led me to:Quote from votdev
And why do you try to modify system relevant code then??????? Does this error output hurt you? It does not stop OMV nor other python code from working, it's only a cosmetic issue.
So I will just ignore it.
I'm getting strange output after every single installation of updates, plugins, packets.
Simple installation of unrar looks like this:Code
I mean last 8 lines. Those are showing everytime I install/update something and I have no idea what do they mean.
OMV is fresh install, after basic configuration and adding MySQL plugin, Docker and ResetPermissions.
Anyone knows what does it mean and where it came from?
I was getting same messages (along with "cannot create socket to [localhost]:2812 -- connection refused") on previous OMV installation but was planning to replace OS drive and reinstall OMV so I didn't care but now I do
Can you help?
Looks like some of your unraid settings were copied among your files. That's why you were seeing this 91 owners, those were remains of previous acl's.
Can you see your files from shell when logging over ssh or if you boot some live CD like Ubuntu?
If yes I would just create new folder with default permissions and then copy all your files to the new directory. Then set any permissions you need using OMV.
I've removed 'noexec' from OMV_FSTAB_MNTOPS_EXT4, mounted drives again and everything behaves the same way it did before.
I'm not even sure is it possible to make new files with 777 permissions by default...
Oh this makes things clear! Had no idea that OMV is using noexec parameter by default.
I found your answer in some older thread saying that removing noexec from OMV_FSTAB_MNTOPS_EXT4 should be enough.
Hope that is also correct for OMV 4.
Thanks Aaron, again. You saved me a lot of time
The one I ordered (some more info) was below 7 bucks (shipping included). I personally can only recommend adapters based on JMicron JMS578 (since supporting UAS, SAT and TRIM) while VIA VL715/VLI716 and ASMedia ASM1351 should be also ok (same feature set but I've only tested a VL716 enclosure and that sucked but no idea whether the USB-C cable was the problem or the enclosure).
Something with even 2 JMS578 is this little board here: aliexpress.com/store/product/O…/1553371_32789632568.html
Following your advice I've ordered an USB internal cable to connect to mobo usb header
and sata to usb adapter based on the chip you suggested (this one http://www.orico.cc/goods.php?id=6358)
Choosen regular Sata instead of M.2 because I have a spare SSD laying in a drawer which can be used for that purpose. Will see how it will work. Thanks again for your advice!
Makes always sense since due to reducing Write Amplification a lot (even if nobody seems to care about this).
But you said you have 'my OMV, MySQL database and home directories' on the pendrive so flashmemory plugin will only take care about a small fraction of your writes anyway...
Using the flashmemory plugin is ok in all cases. I even use it on my systems with SSDs.
Flashmemory plugin installed and running. Thanks!
Samba uses the last parameter that it reads. So, 21 and 22 will be the lines it uses. That said, to change the default, add the following lines to /etc/default/openmediavault:
Thank you Aaron! Added those lines and it's working exactly how I wanted.
Aaron one more question about permissions if you don't mind.
For newly created files (from ssh, downloaded through Transmission etc.) permissions are being set to 666. Any way to change this?
I have already set ACL's for that folder as follows:Code
But still whenever I create a new file inside 'folder2' it gets 666 permissions instead of 777. No idea why...
Can anyone help?