New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
@William
I largely agree (and compliments; your off the top of your head hw knowledge is amazing). It just seems that you are more hardcore in some respects than myself.
But look, where you see "no eec? crap!" and were right in an enterprise or DC context, I see "Great! He has finally raid 1'd disks!". There are many consumer minded people out there ...
$ dmesg | grep -i cpu cpu0: Intel(R) Celeron(R) M processor 1.50GHz ("GenuineIntel" 686-class) 1.51 GHzThis single-core monster out of 2005 (M370 I think) drives a pair of SATA drives' nfs without much issue...on OpenBSD softraid even, the poor thing. Not saying I'm streaming 4K to a dozen iPads or anything but as you observe, the CPU isn't likely to be the first bottleneck.
Anyone with DIY builds having lasting success using btrfs in place (or on top) of standard Linux mdadm for their NAS applications? Been reading about btrfs implementation of RAID1 on mismatched disks, while still having some checksumming capability available. Looks like it's becoming more stable for level 0 and 1 implementations...not so much for 5 or 6. Any gotchas or complete disasters on home/soho applications?
I've got pair of 3TB Seagate drives in individual UAS capable USB3-SATA enclosures in software RAID1, and a pair of 2TB refurb Hitachi Ultrastar drives in a dual-bay USB3-SATA dock in software RAID1 being served over NFS by an Orange Pi PC2 using a USB gigabit ethernet adapter (note OPi PC2 is only USB2.0 capable). I can get 38MB/s+ out of the 3TB array, and the 2TB refurb drives array I'm getting far less. The OPi PC2 is also my plex server. Feeds two direct play 720p streams to Roku's in the house without issue.
Edited to rephrase the question. Edited again to add my current NAS setup
My disks are mismatched, I use a software raid of partitions.
Most of my access is from mobile devices over wi-fi so the slowest disk has plenty of speed for this. There have been cases of powerloss before i fitted it with an UPS and a few disks have been trashed before, but were already in a very bad shape in the moment i added them in the array. I could recover all data, each time.
If those are ST3000DM01 drives consider replacing them or having a decent backup (That model is rather unreliable)
Indeed. I purchased four of them over various months. When I eventually repurposed them for my DIY NAS I had three of them fail in the first six months.
I recommend anyone using them regularly monitor SMART status and scrub the drives for read errors; while having in mind a plan to replace them.
BTRFS Raid5/6 really isn't as terrible as people seem to make out, or maybe i've just been really lucky.
Also most of the issues are supposedly fixed in kernel 4.12 yes it still has the write hole but that isn't unique to BTRFS, could use checksum support for parity but it's no worse than a lot of other raid's that don't checksum parity either.
Maybe not the thread for it, but I'm curious because I keep reading nothing positive about btrfs unless it's used in mirrors. How long have you been running it? What kind of load? Any disk failures? Etc.
Fairly light load mostly, Was created in rockstor originally but manually imported it into OMV, disks passed through on a proxmox host - was a C2750-d4i before the board/cpu died
had a HDD develop bad sectors and replaced it, also had a drive that I accidentally wiped during a proxmox re-install (was doing it over IPMI or i'd have pulled the drives to stop that being a possibility). It's also been re-balanced several times, at one point to Raid1 when the Raid5/6 bug became known and recently to raid 6
As far as I know even when raid5/6 was buggy I've not lost any data to it, one of the reasons for removing it from rockstor was it requires quota support and doing anything like a large snapshot delete or re-balance with quotas on causes massive load.
Now I'm not saying BTRFS is the be all and end all, but the reason I started using it for my NAS/Storage server was because you can re-balance it to add/remove drives.etc and I've not really seen any other filesystem easily offer that.
I would point out if It's data I care about I don't trust it to any single raid, it's backed up to multiple machines