Posts by olduser

    Just turn off e-mail notifications.


    Alternatively you could setup a script with your desired logic on OMV to respond if all is O.K. to your liking. eg. Client: "Hey there sexy!", OMV: "Back at ya!". I mean if you _REALLY_ cared, you would do this anyways. I don't think Spetsnaz is targeting me, but I still have a script to check rsync logs for my pictures.

    ...make sure to wear your tinfoil hat.

    How did he KNOW you have a tinfoil hat!!!! Spies everywhere!!!

    Potentially, yes! I was directed to this thread from another where I had inquired about a pending change email I'd gotten. Because I knew I had not changed anything, klaxons and huge flashing red lights started going off in my head. If the system wants changes committed and I haven't made any changes, then I can only assume Very Bad Things are afoot! Good to know that it's benign, but it's very alarming.


    That said, I do really like OMV and it's served me well for years now. Keep up the good work & thanks!

    The prior context was exclusively for unnoticed mouse clicks. In your case it seems that everything is good. Now when you've found your system reconfigured and you DIDN'T get an email...

    don't understand the requirement ECC (which is the limiting factor) because you are building a single user machine with a small RAM size.

    ECC comes into play with terabytes of RAM

    ECC is not a cure all just like RAID isn't a backup but, ECC is a pinch of prevention just like RAID.


    If you're not storing irreplaceable files, ECC is near useless for home NAS users. I'm a active photographer so ECC is taken for granted and I use all the paranoid ECC options.


    NOTE: These Ryzen CPU's don't really _SUPPORT_ ECC, they tolerate ECC's single surface capability. So if you want mirroring, scrubbing, etc. they won't do. I really don't know the innards of Ryzen ECC functionality so I don't know if it is CPU or motherboard limitations, but of the 20 or so motherboards I looked at, none of them had configurable ECC options.

    This isn't in a rude tone, but if there exists a platter drive you'd think will suffice for home NAS, then CMR or SMR shouldn't be factored. Also, a home NAS with enough RAM can negate nearly all OS writes to platters using virtual drives.


    In my opinion, modern large platter drives == "the new tape", although film is probably more reliable in the long run, but definitely not cheaper to store (but to be fair I haven't read a store manual for film in 20 years, so maybe it's better).


    Of course, platter drives are also affect by humidity very similar to camera lenses, which is too much == fog/moisture. Granted this is old person voodoo (probably FUD), but if a platter never stops spinning, then moisture can't build up. It something rarely talked about, but not everyone lives in Arizona (in the Philippines it's an absolute concern).

    What is that kind of "Game ROM/Audio Jukebox" Project? :D


    Greetings

    David

    It is or was or can be a Ui entirely for kids < 10 that can launch games/music. I finished the Ui in conception of what a kid would use (still needs the meta data junkie stuff for parents), but then back in node when I went to finish it up, I thought to myself what about a kids jukebox?... it's one of those projects.


    The string speed of node makes it addicting to write text parsers, but the Ui is still what interests me the most as there seems to be jack shit for kids (really... there's nothing I've seen). From an adults perspective, the Ui makes nearly no sense so I'm proud of it :-).

    FWIW, with the GPU you shouldn't need a PCIe 3.0 x16 header, just a x4 header. You might be able to get away with a x1, but if not (and since a x2 doesn't exist), a x4 will work. Even if you use a x4 it might be worth trying an x1 adapter in the x4 to keep lanes open.


    I know the Nvidia 1660 Ti works in a x4 for decode/encode. Also, as you seemed to be using a desktop board and want to keep the x16 header for something else (HBA or 10GB), M.2 adapters exist to break out to a x1 or x4.


    Here's a 5 pack of M.2 to x4 with the header open (so you don't have to cut it)

    https://m.aliexpress.com/item/…art-amp.item.32830559009&

    While I agree with macom, I'm going to play devil.


    If the abstraction can't be jailed in a tidy manner, I don't use docker. I'm working with a homelab for sensor data and while it's technically "containerable", it becomes curiously buggy due to various system comms/protocols. For instance 1 of the 2 I've tried didn't like SPI and the other didn't like multiple devices on the I²C (although I think that was a parser/software limitation).


    Containers are also a nightmare for any development work at all as adding even a simple custom syntax colouring file can be frustrating, never mind full modules and this could possibly extend to "office" programs as well (not sure though).


    Then there is performance. While you can run a web server in a container on a fancier computer, running one in a container on a cheap VPS/host isn't a great idea and if you had a super-awesome web server, it wouldn't make sense either. Also, I have this ever-running... never-ending... web based "Game ROM/Audio Jukebox" project I started in node.js (4 years ago :-/) that when parsing audio data or hashing checksums it creates such a huge variance in RAM and core usage that there's no real way to scale that in Docker that makes Docker "worth it" (or any containers really).


    But yes, containers are great when you just want to try something out or the amount of configuration to the system could be considerable (I still have Mailu and Gitea under Docker for those reasons... Mailu especially is kind of a beast).

    2 months ago I learned Docker and nothing made Docker more understandable than using docker from the shell. Portainer hide away far too much and compose, while nice, changed too much syntax to match/meet YAML rationality.


    With the above said, the first tutorial I used was macom's compose tutorial from which I switched to the official docs. I couldn't imagine starting with Portainer... that would be frustrating (to learn, it always seem better to go down rather than up).

    I'm not trying to step on toes here, but for the top piece, why not recess the button bindings all the way around the bottom and remove the need for supports? Or, print the entire thing as 1 piece and create a rectangular button lid allowing the drives to be entered in a swapping fashion from the side , the finger hole to pull the lid could also be used as intake/outake. You could also replace the button bindings with ribbed bindings, this would allow thicker layer heights with reliable dimensions.


    nanopim4-cgomesu-case02.jpg


    I think it might be worth mentioning to not print this with PLA. PETG seems like the cheapest and safest material to use to negate future climatic deformation. PLA is great for lightweight functional parts, but so long as you plan to replace them (and dehydrating PLA won't help this). Also, for that exact printer, it might be worth measuring the X and Y axis as the layers appear to be shifted (and possible under extrusion). Or if cosmetics isn't a concern, bump the extrusion width up a bit and double the layer height (the quick and dirty guarantee).

    Also worth a look in Mini ITX is the https://www.asrockrack.com/gen…3558D4I-4L#Specifications


    There is a concern about how many drive ports are actually available though. See this review:


    https://www.amazon.com/ASRock-…GX8SFKATQ#customerReviews

    Here's another review stating the same, it seems like a bad reference board for NAS...


    https://www.newegg.com/asrock-…ore-16w/p/N82E16813140014


    Also the manual supports both reviews in regards to this aspect. That board might be the most deceptive board on ASROCK's website. I understand the limitations but, when you see 2 mini SAS HD ports you can see some kind of allurement that's really a gotcha.

    Aluminum to bend up? I don't get it. I gave you my complete parts list. You need a PC case with fans to hold the drives and power supply and the data interconnect cables.


    What you are going to come up with lacks any consideration for drive cooling and will leak a lot of RF in and out.

    I am taking your suggestions on EVERYTHING BUT the case... surely you can understand I want to tinker a bit myself :-) BTW, if I didn't already have a NAS of any type, I would undoubtedly buy that so that link is still good to pass forward and I still might buy it (but $200 for just a drive enclosure, I have to at least attempt DIY first). Honestly, thank you!

    What is this "SATA 4-in-1 cable" you mention?


    Perhaps you are missing the point that all the cables, LSI cards, and bulkhead feedthru card that I used are both SAS and SATA compatible.

    I think I bookmarked the 4-in-1 SATA cable at home, but it seemed horrible as it literally looked like someone just put mesh around 4 SATA cables and molded up proprietary headers (they came in vertical and horizontal orientation)... I really want no part of it.


    Perhaps I am missing the point that you're doing that if it's not for connection and power stability. I'm not a server person so there is vasts, vasts amounts I don't know, but when I see that the cost is cheaper, connections are more reliable and the user/help base is largely revolving around just a few reference vendors... it kind of all seems like a win/win.


    Anyways, I'm off to make my life hell and buy some aluminum to bend up, I'll take pictures of all my failures.

    ...don't ever buy a used SAS drive.


    My server used to be in the Lian Li PC-D8000 case..

    Funny, I thought the only storage medium I've bought used in 25 years was floppies for old PBX style systems, and the first image that loaded for the D8000 looked like an old PBX... a moment of irony for me :-)


    But yeh, I'm not buying SAS new or used for this exact thing. Now the power distribution is why I'm on the fence about the backplane. On one hand I don't want it, but on the other the wire mess (of course I won't have 32 drives to support the directory for the Eastern region :-P).


    But damn, I just wanted to bend up some aluminum in a DIY fashion and say HA! $50 bucks!!... but that's rrrrrrreally not looking plausible.

    It shouldn't make a difference whether you use SATA or SAS. The price of SAS drives will be much higher though.

    Actually this is for SATA drives. Granted, I could be approaching this all wrong, but it appears that *IF* someone is willing to buy used, then the price of all this is much, much cheaper using SAS components than SATA, at least for adding more than 4 drives (adding just 4 drives then maybe not).


    I bought one of these for $20 USD, it works flashed fine... Lsi-Sas-9200-16e-Host-Bus-Adapter


    The SATA 4-in-1 cables (that I can't find the link for right now) were $20 a piece, shipping wasn't free, I couldn't find them used and their length was only 20 inches max, not to mention I'd need 2 of them pushing the price to a realistic $50. However, if one is willing to buy used (I am), I can find 2 SAS cables for less than $15 shipped. Now, the breakout cables are still needed, but at that point those would always be needed (unless you used a REALLY long SATA cable and connected board to board).

    Why SAS drives?

    Honestly, cable management for a secondary box.


    My OMV server is in a Silverstone DS380 case https://www.silverstonetek.com/product.php?pid=452


    When I added the eighth and final drive to it I began planning for a seperate DAS box to accommodate more drives in the future.


    I added an LSI 9200-8E SAS card to ....

    Right, right right right. You're at where I'm trying to get to, but when it came to adding the 2nd box, did you feel it was too much space for just drives? That's what's hindering me on the 2nd box... the feeling of too much space for just drives. However, I might have to go that route just to fit a PSU as I'm having a surprisingly hard DIY time finding a lower power PSU without factoring in buck converter none sense for the 5v.

    Looking for suggestions for an enclosure that can contain 4 or 8 drives with a PSU and SAS header. I can't find one that is just that and nothing else. I'm on the fence in regards to backplanes as I don't plan on swapping out home NAS drives often.


    For the structure itself and apart from bending aluminum, I see 5.25" drive cages that can hold up to 5 @ 3.5" drives. Outside of these, I haven't found any type of enclosure that can be cut up to meet these needs (at least not affordable options).


    The SAS header/adapter seems to be a problem as I can't find gender adapters, ie. female to female. Well, there is this but it has a PCB Linkreal 2 Port SFF 8088 to Internal SFF 8087 Transfer Card sff 8087 to sff 8088


    Any suggestions welcomed.

    If a power supply has 2.5a in standby at 12V, why would it need that much on 12V? I thought 3.3v at like ~0.3a was for standby, but now I'm curious as 2.5a seems like a lot, especially on the 12v.


    With OMV, if I put X sata drives in a box, would such a psu stay in standby if only 1 drive was active and say 7 are sleeping? The main reason I'm asking is that some of these psu's can be pretty loud... but in standby...


    Also, can SSD's get their 5v from the 12v bucked? I know an ATX gives 5v across the 5v obviously, but I've never tested an SSD to see if it can use the 12v if 5v isn't present.

    I'm sure whatever you're using is fine, but make sure that whatever you use keeps/gives proper permissions. I've never worried about this as what I've used is for Arduino, but I did notice that by default on a proper desktop, all files uploaded belong to which ever user the webserver was (logically).


    It's rare, but some countries (ie. India) have text encodings that can be unexpectedly tricky, which normally isn't a problem, but since this text could be a filename... something to think about at least.


    Lastly, always make sure that you don't already have URL support for sftp:// (I'd even take FTP over any HTTP).