|
app backup just saved my rear end this week when btrfs did it’s thing and corrupted
|
# ? Jan 30, 2023 04:34 |
|
|
# ? Jun 3, 2024 22:31 |
|
got everything from the beelink gk55 migrated to the ML30, so far so good, once the parity finishes syncing I’m going to move the plex container to it from the gen10+ so it can take advantage of the p400 dug out my old stack of 4tb drives to replace the single 12tb external I had on the gk55 it lives in my shed as a faux offsite, it’s currently 0°f outside and the processor is idling at 5°c
|
# ? Jan 30, 2023 05:03 |
|
Flyndre posted:I've set up OneDrive to sync to my NAS using Synology's Cloud Sync package. I plan a workflow where I might keep working files on my Mac in OneDrive (photos due to be edited for instance), before I transfer them to a mounted SMB folder on the NAS. Hyper backup: definitely no interference. Hyper Backup is de-duplicating (via file hashes), so as long as the process of moving them between your Mac, the OneDrive cloud, and the NAS does not modify them* only one copy of the data will be kept. *No guarantee that said process will not modify files. Some apple software has in my not-at-all-recent experience been lovely about loving with metadata without asking. NAS snapshots: I think yes, it will store data twice in snapshots. Only expensive high-end units support de-duplication on the drives apparently. I doubt this is a big enough deal to care about.
|
# ? Jan 30, 2023 05:49 |
|
Is my processor load on this ml30 supposed to be routinely between 80 - 100%? What's the most common culprit?
|
# ? Jan 30, 2023 06:24 |
|
No. Have you run "top" to see what's using so much cpu? Assuming you're on *nix.
|
# ? Jan 30, 2023 07:28 |
|
Captain Apollo posted:Is my processor load on this ml30 supposed to be routinely between 80 - 100%? What's the most common culprit? it’s a 7 year old dual core processor while the E1270v5 is just as old, it’s at least a quad core with hyper threading, and a basically free upgrade is your parity drive still syncing?
|
# ? Jan 30, 2023 07:57 |
|
Weird it's all quiet on the western front now. It was doing some serious hard work earlier but seems to be good and normal now.
|
# ? Jan 30, 2023 08:30 |
|
$7 key from ebay to get full ILO access? don’t mind if I do
|
# ? Jan 30, 2023 09:10 |
|
e.pilot posted:$7 key from ebay to get full ILO access? don’t mind if I do What benefits does ILO access provide? I pretty much disabled it in the BIOS as soon as I received the machine. E: also which key did you buy? For $7 I'd like to tinker with it too.
|
# ? Jan 30, 2023 15:15 |
|
Double post because my HDDs come in tomorrow. What's the current standard for initial testing for hard drives before setting up in a NAS? I see lots of talk about burn in when I Google it but I can't find much talking about the actual process.
|
# ? Jan 30, 2023 15:27 |
|
SpartanIvy posted:What benefits does ILO access provide? I pretty much disabled it in the BIOS as soon as I received the machine. I have a stack of servers sitting behind me that I have fully set up the BIOS, installed the OS, and configured the software having never connected anything more than one power cable and one ethernet cable. For a home NAS it basically means you can shove the thing wherever it's convenient without having to worry about easy access to connect devices. If the thing is sitting right next to your main PC it might not matter to you.
|
# ? Jan 30, 2023 15:51 |
|
SpartanIvy posted:Double post because my HDDs come in tomorrow. What's the current standard for initial testing for hard drives before setting up in a NAS? I see lots of talk about burn in when I Google it but I can't find much talking about the actual process. Write some data. Do a speed test with crystaldisk or atto to see that performance is as expected and fly the heads around a bit. If you want to be super-thorough you could do a surface scan. IMO a burn in like doing a zero write to the whole drive is useless. Backblaze stats don't show a bathtub curve of high 1st-year failures in drives these days, drives pretty much are either DOA or work. So the idea that you could weed out a bad drive with burn in seems unlikely.
|
# ? Jan 30, 2023 16:42 |
If you're running something based on FreeBSD, diskinfo -cit on a disk that's not got any data written to it (preferably a disk that's completely uninitialized) can be quite revealing. Especially if you have an idea of how it should perform, from having run it on similar drives that you know are good.
|
|
# ? Jan 30, 2023 18:45 |
|
SpartanIvy posted:Double post because my HDDs come in tomorrow. What's the current standard for initial testing for hard drives before setting up in a NAS? I see lots of talk about burn in when I Google it but I can't find much talking about the actual process. I did a burn in of my 18gb Exos from Server Part Deals. Generally I saw using badblocks but it's REAL old and couldn't get it to read my drives. I think if you mess with the block size you might be able to but the commands I saw didn't work (mostly here). I ended up going with this command from Arch Wiki of running the following commands code:
code:
|
# ? Jan 30, 2023 20:35 |
|
SpartanIvy posted:What benefits does ILO access provide? I pretty much disabled it in the BIOS as soon as I received the machine. HP’s flavor of IPMI, in other words a PiKVM but built in e: sorry it was $8, lol https://www.ebay.com/itm/254154313587 also got the p400 up and running, 4x4k transcodes without missing a beat, I ran out of things to stream to, this thing is a budget beefcake e.pilot fucked around with this message at 22:25 on Jan 30, 2023 |
# ? Jan 30, 2023 22:22 |
|
Klyith posted:Write some data. Do a speed test with crystaldisk or atto to see that performance is as expected and fly the heads around a bit. If you want to be super-thorough you could do a surface scan. I think SpartanIvy is getting a batch of refurb drives anyway. I run nwipe on any new-to-me drives with a DoD Short test, most failures that I uncover this way show up within the first pass - often pretty early because throughput will drop to a crawl.
|
# ? Jan 30, 2023 22:33 |
|
SpartanIvy posted:What benefits does ILO access provide? I pretty much disabled it in the BIOS as soon as I received the machine. iLO access can be quite useful, especially for the Integrated Management Log and console access. For TrueNAS and Unraid use the iLO Advanced license isn't that necessary, HPE allows console access during POST for configuring BIOS and devices. But with those eBay prices might as well get it. Of course if you enable iLO remember to to update it regularly. https://support.hpe.com/connect/s/softwaredetails?language=en_US&softwareId=MTX_a9afd91f006a44f8b1b5d0e09b Updating is a bit tricky. Extract the "ilo4_281.bin" from the .exe with 7-Zip and upload it through the iLO, "Administration - Firmware". Update the BIOS to at least version 2.82. It's a critical update, so downloading it doesn't require a support agreement. https://support.hpe.com/connect/s/softwaredetails?softwareId=MTX_ee67dcd89ef74e2da195fdff53&language=en_US Other hardware components have firmware updates too, but I think most of them can't be updated through iLO, so they would be more tricky.
|
# ? Jan 30, 2023 23:59 |
|
Saukkis posted:iLO access can be quite useful, especially for the Integrated Management Log and console access. For TrueNAS and Unraid use the iLO Advanced license isn't that necessary, HPE allows console access during POST for configuring BIOS and devices. But with those eBay prices might as well get it. Thanks for this. I knew there was a BIOS update available but hadn't gotten around to installing it. I didn't realize we were past the days of using bootable media to apply firmware fixes. Using a web interface from my PC feels like the future.
|
# ? Jan 31, 2023 01:35 |
|
SpartanIvy posted:Thanks for this. I knew there was a BIOS update available but hadn't gotten around to installing it. I didn't realize we were past the days of using bootable media to apply firmware fixes. Using a web interface from my PC feels like the future. Well bootable media could be convenient, but you would need to get your hands on Service Pack for ProLiant and that requires support contract or new hardware. If you have an RPM-based distribution you can probably install the firmware from OS. If neither applies it may be tricky. My solution for updating Dell servers with Ubuntu installs was to install CentOS 7 on a USB stick and run Dell System Update from it.
|
# ? Jan 31, 2023 02:38 |
|
Well, if you put a PCIe card in an ML30, it's going to complain about missing the front fan and then if you disable that warning it will run the other fan at 100% perpetually. Fun.
|
# ? Jan 31, 2023 05:51 |
Saukkis posted:Well bootable media could be convenient, but you would need to get your hands on Service Pack for ProLiant and that requires support contract or new hardware. If you have an RPM-based distribution you can probably install the firmware from OS. If neither applies it may be tricky. My solution for updating Dell servers with Ubuntu installs was to install CentOS 7 on a USB stick and run Dell System Update from it. One exception appears to be the Microserver Gen10+, where you need an active service contract in order to download the firmware. AlternateAccount posted:Well, if you put a PCIe card in an ML30, it's going to complain about missing the front fan and then if you disable that warning it will run the other fan at 100% perpetually. Fun. HPE is special though. My HPE DL380p Gen8 will run the fans at 30% with no daughterboards inserted, and 30% if it's a HPE branded daughterboard. If you insert a daughterboard that's not HPE branded (even if it runs the exact same firmware as a HPE branded card), it'll boost the fans to 60%).
|
|
# ? Jan 31, 2023 13:21 |
|
BlankSystemDaemon posted:I snipped this because most of it has been addressed, but just want to add that the key term you're looking for is: JBOD DAS - uninitialized, it means just a bunch of disks (via) direct attach storage. Ah yes, sometimes it's just not knowing the terminology I am looking for. So what I ended up doing it getting this OWC Mercury Elite Pro Dual with 3-Port Hub and then bought 2 of these Seagate Exos 12TB renewed drives on Amazon. No problem getting both to show up as individual drives and using Time Machine to back up one to the other. Still a little skeptical about getting renewed drives, but the whole thing only cost $400. Since it's DAS and I can mount Dropbox on it, I plan on using Dropbox to backup my most important data on there (home videos/photos), and not backup Plex Library type media. So worst case scenario if both drives die at the exact same time I just lose my Plex library, which is no big deal. And then my most important data follows the 3-2-1 rule. So I think I'm good!
|
# ? Jan 31, 2023 17:27 |
|
Pardon my newbie question; there's just a lot of information in the thread and I've tried searching for a bit but couldn't find a recent answer. I currently have an Ubuntu PC with a a Plex install where I I create a Plex username for family to log in and view media. I'm interested in expanding that to an entry-level NAS with Plex still being the main use case. Looking at either the Synology DS420+ or QNAP TS-464. I see that both Synology and QNAP have histories of being hit with ransomware attacks (QNAP as recent as Aug 2022). Is there any way to somewhat securely open up the server to outside-the-house Plex streaming or is this just like a don't loving do it kinda thing? Would compartmentalizing the data/users in some way work? If that can be safely done, I'd also be interested in using the NAS to back up ~2TB of personal data, but that wouldn't be the main use case. I wouldn't care about accessing the PC backup items outside the house. Has my dumb rear end just been lucky keeping a Plex server running raw on my Ubuntu box without issue?
|
# ? Jan 31, 2023 19:40 |
|
What's wrong with your current Ubuntu/Plex setup? IMO that is a superior config to a proprietary nas. Regarding outside access to plex, the recommended solution is a reverse proxy or a cloudflare tunnel. Though, cloudflare doesn't want you streaming video on their service. There is some docker for reverse-proxy on nginx that I keep seeing mentioned, but I can't recall the name.
|
# ? Jan 31, 2023 20:32 |
|
I was under the impression (that may be entirely wrong), but if you invite a plex user to your library they handle all of the tunneling and stuff and you don’t have to configure that nor open your device to the internet. If I have to do a bunch of network janitoring I think my brother is about to be real bummed lmao
|
# ? Jan 31, 2023 20:56 |
|
Comatoast posted:What's wrong with your current Ubuntu/Plex setup? IMO that is a superior config to a proprietary nas. It's not too great of a PC and wasn't really originally constructed with the forethought to be a media server. It is a a 6 year old Intel G4600 inside of an ASRock Deskmini 110W. The Deskmini has an known issue with integrated graphics which causes it to freeze occasionally. I've been unable to find a fix for it; sometimes it'll stay up for 3 weeks with no issue and sometimes it'll crash 3 times in one evening. It kind of evolved into the family server since I kept it plugged into to my TV due to the small form factor. I'd like to get something more stable and repurpose it for other tasks that won't annoy me as much. If I could go back in time I'd probably wait for another revision but otherwise I love the form factor. FAT32 SHAMER posted:I was under the impression (that may be entirely wrong), but if you invite a plex user to your library they handle all of the tunneling and stuff and you don’t have to configure that nor open your device to the internet. I thought so too, but then I started reading about the malware & how/why to set up docker & my brain went uhhh what am I getting into.
|
# ? Jan 31, 2023 21:01 |
|
I've had good luck with Plex working almost entirely out-of-the-box on a few installs (desktop, TrueNAS Core, TrueNAS Scale) - but it'll depend on your router, as long as it supports UPnP you're probably fine. Plex is also nice that it reports in settings if it's accessible from the internet-at-large, makes it easy to troubleshoot.
|
# ? Jan 31, 2023 22:28 |
|
There are a couple decisions that will guide your nas journey: how many drives do you need, do you insist on ZFS, and do you mind janitoring your own general purpose OS. Personally, I find Ubuntu to be just as set-it-forget-it as a synology. You’re going to have to learn how to configure something, may as well be linux software instead of the proprietary synology gui. I run a HP S01. It can only hold a single 3.5” drive, but that’s all I need. I did throw 16gb of memory, a cheap nvme drive and an Intel 10400T processor in it. Roughly $300 all in. If Dell or HP ever have an excellent deal (like they were doing around 2013) on a micro-atx tower with ECC memory, then I’ll upgrade to that. Comatoast fucked around with this message at 22:57 on Jan 31, 2023 |
# ? Jan 31, 2023 22:32 |
|
Trapick posted:I've had good luck with Plex working almost entirely out-of-the-box on a few installs (desktop, TrueNAS Core, TrueNAS Scale) - but it'll depend on your router, as long as it supports UPnP you're probably fine. Plex is also nice that it reports in settings if it's accessible from the internet-at-large, makes it easy to troubleshoot. Isn't keeping UPnP on a much bigger security risk for your network than forwarding Plex's port outside your network?
|
# ? Jan 31, 2023 22:40 |
|
Corb3t posted:Isn't keeping UPnP on a much bigger security risk for your network than forwarding Plex's port outside your network? edit: ok, I'll stand slightly corrected - QNAP especially seems to have a bunch of weird services opening up holes and getting wrecked, be very careful. Trapick fucked around with this message at 23:23 on Jan 31, 2023 |
# ? Jan 31, 2023 23:07 |
|
FAT32 SHAMER posted:I was under the impression (that may be entirely wrong), but if you invite a plex user to your library they handle all of the tunneling and stuff and you don’t have to configure that nor open your device to the internet. Plex does have an option for this, called Plex Relay. The downside is that quality is extremely limited, 1Mbps if you don't have Plex Pass, 2Mbps if you do. The much more sensible option is to simply open port 32400 TCP to the internet, without exposing anything else.
|
# ? Jan 31, 2023 23:17 |
It's not exactly difficult to setup headscale which lets you self-host a beacon for tailscale. It's even available in FreeBSD Ports so can be installed in a jail.
|
|
# ? Feb 1, 2023 00:10 |
|
Astro7x posted:Ah yes, sometimes it's just not knowing the terminology I am looking for. I'm nearly complete with a 4x4tb to 5x10tb drive migration using all refurb drives. I like it so far, 3 out of 4 had 0 hours, one has 2.5 years and the last is in transit. I staggered purchases over about 4 months. The cost is so much less that I'm willing to play the numbers game with them. I think I'd prefer replacing a dead refurb compared to dealing with consumer level RMAs. I'm not saying those 0 hour drives are in fact accurate values, but they've been working well so far. Dyscrasia fucked around with this message at 00:55 on Feb 1, 2023 |
# ? Feb 1, 2023 00:50 |
|
BlankSystemDaemon posted:That's fairly normal, unfortunately - even Supermicro and TYAN will up fanspeed if one fan is missing and there's a daughterboard inserted, because the daughterboards depend entirely on the airflow of the fans to keep the components cooled. Yeah but it suuuuuucks. Hopefully whoever's testing that $5 Delta gets a good result.
|
# ? Feb 1, 2023 00:55 |
|
Dyscrasia posted:I'm nearly complete with a 4x4tb to 5x10tb drive migration using all refurb drives. I like it so far, 3 out of 4 had 0 hours, one has 2.5 years and the last is in transit. I staggered purchases over about 4 months. The cost is so much less that I'm willing to play the numbers game with them. I think I'd prefer replacing a dead refurb compared to dealing with consumer level RMAs. How do you check the hours on a drive? I am on a Mac if that matters
|
# ? Feb 1, 2023 06:53 |
|
IOwnCalculus posted:Plex does have an option for this, called Plex Relay. The downside is that quality is extremely limited, 1Mbps if you don't have Plex Pass, 2Mbps if you do. Oh, I thought all that required janitoring like this: BlankSystemDaemon posted:It's not exactly difficult to setup headscale which lets you self-host a beacon for tailscale. So instead of using relay, your friends have to connect to a VPN that you’re hosting for your network to then be technically on the same network and not be rate limited to 1-2Mbps? I didn’t read all the documentation yet, but that sounds like a lot of effort unless this lets them stream at whatever your upload/their download speed is What does port 32400 TCP do? Does it somehow surpass the 1Mbps from relay, or is it an easier way for them to do the same as a VPN? Sorry for the somewhat off topic posts, I’ve avoided network janitoring since it didn’t really benefit me and had a whole lot of ways to harm me so a lot of this I’m just learning
|
# ? Feb 1, 2023 07:04 |
|
Plex Relay means all traffic is going through one of their servers (the relay), they limit the traffic because they don't want to be paying for all that bandwidth. Either the stream goes server->relay->client or server->client; if the second, you need to either be on the same network (physically or virtually) or have a port open somehow. Opening a port (e.g. 32400) is easier than setting up a VPN, yah. On your side either is probably fine, but your friends will have to do nothing if you open a port (they login to Plex, you've shared access with them, bingo bango) or go through a pain in the rear end for a VPN. If they're watching on a computer minor pain in the rear end, if on a Roku stick or xbox much more of a pain. Open a port, keep plex updated/watch for news of exploits, don't worry too much about it. edit: oh and wherever Plex is running, limit it - have it run under a user with minimal permissions, read-only access to your media, write access to its own metadata and stuff only, etc. Trapick fucked around with this message at 07:35 on Feb 1, 2023 |
# ? Feb 1, 2023 07:30 |
|
BlankSystemDaemon posted:It's not exactly difficult to setup headscale which lets you self-host a beacon for tailscale. I feel like we're getting close to the point where it'll be possible to buy a cheap device that runs a wireguard endpoint and send it to your non-technical friends and family, tell them to plug it in, and it automatically bridges to your network and everything is super easy
|
# ? Feb 1, 2023 09:00 |
|
Klyith posted:IMO a burn in like doing a zero write to the whole drive is useless. Backblaze stats don't show a bathtub curve of high 1st-year failures in drives these days, drives pretty much are either DOA or work. So the idea that you could weed out a bad drive with burn in seems unlikely. It's a waste of time mostly. Most drive failure modes are not going to be accelerated by just doing a bunch of write operations. A drive that fails after however many days of whatever "stress test" you put it through is in many cases a drive that would have failed by that time anyway. And just writing to every single sector isn't going to cause a drive destined to fail in 6 months to instead fail during your test. The most that can be said about tools like badblocks is that they may uncover SMART errors that otherwise wouldn't be apparent until you've filled the drive with data. But a large fraction of drives that fail do so without ever having any SMART errors...and SMART error signals themselves don't have great predictive value for drive failure, especially for individual drives. Like, the most common read/write failure modes for HDDs are (in roughly descending order, and excluding extrinsic things like vibration or shock): 1.) Adjacent track interference--where (many) repeated writes to a given sector degrade the pattern written on adjacent tracks. Fundamentally this is a thermally activated process and is not going to be accelerated by just writing to every sector of the drive like badblocks does. 2.) Reader or writer degradation or instability. When not caught by in-factory tests, these are often a consequence of media or head defects that manifest simply over time, like corrosion or tribology problems. They will cause a failure when they do, and just writing a bunch of data won't speed things up. 3.) Unrecoverable defect on disk--again, if not caught during factory self-test, this is often the result of a slowly-building corrosion or contamination problem. 4.) Off-track write. The writer wasn't able to follow the track correctly when writing. Often caused by a poorly written servo or degradation to the servo pattern. Nothing you do in software can read or write to the servo pattern, so, again, just writing a bunch won't accelerate these problems At their core most of these are "ticking time bomb" problems--which is why they are hard to catch during factory test and therefore why they are the most common. If you really want to do an accelerated stress test, just writing to every sector is pretty pointless. Running at sufficiently elevated temperature *is* an effective accelerated stress test, but not advisable since it will reduce the lifetime even of perfectly good drives. Anyway thats my ted talk
|
# ? Feb 1, 2023 09:33 |
|
|
# ? Jun 3, 2024 22:31 |
|
Basic burn in tests are mostly just suitable to find DOA drives. I've experienced one DOA drive since I started building computers in the 90s. My seagate lp based array had failures when temps got high, that's something others have documented as well.
|
# ? Feb 1, 2023 09:40 |