Xpenology expansion unit?

[Sdeve]

Firstly, I have managed to install an Xpenology DSM7.2.1 and it seems to be running just fine, so a big thank you to whoever provided the means to do it.

 

I have a couple of Synology NAS units, one being a DS920+. I've been looking at expansion units which has got me wondering if there is lurking somewhere an Xpenology setup to build an expansion unit for a 'genuine' Synology NAS. It strikes me it would come in handy maybe?

[Sdeve]

Anybody?

 

[DSfuchs]

I expand real devices and Xpenologies via external USB3 enclosures. These are also available as internal drives via Settings.

[IG-88]

On 10/24/2023 at 12:47 PM, Sdeve said:

I have a couple of Synology NAS units, one being a DS920+. I've been looking at expansion units which has got me wondering if there is lurking somewhere an Xpenology setup to build an expansion unit for a 'genuine' Synology NAS. It strikes me it would come in handy maybe?

 

no, there is no such thing that tricks dsm (xpenology or original) into accepting a "normal" sata (or sas) enclosure as one of there original expansion units

afaik there is some tinkering in specific drivers kernel code and firmware checks for the external unit involved

(was introduced ~2012/2013 before that any eSATA unit that presented every disks as single disk worked)

also worth mentioning is that technically its a simple sata multiplier and you use the amount of disk in that external unit through that one 6Mbit sata connection, all drives involved (external unit) sharing this bandwidth) as long a you just have to handle a 1Gbit nic you wont see much difference but if you want to max out what lets say 4 + 4 10TB drives can do and use a 10G nic you will see some differences and raid rebuild might speed might also suffer in eSATA connection scenarios

 

also a general problem with that kind of scenario is reliability as when accidentally cutting power to the external unit there will be massive raid problems afterwards usually resulting in loss of the raid volume and when manually forcing repairs its about how much data is lost and hot to know what data (files) is involved

i dont know if synology has any code in place to "soften" that for there own external units (like caching to ram or system partition when sensing that "loss" by a heartbeat from external and bringing the raid to a read only mode to keep the mdadm raid in working condition)

 

as you can use internal up to 24 drives with xpenology and only your hardware is the limit (like having room internal for disks and enough sata ports) there is only limited need for even connecting drives external and some people doing this hat seen raid problems

if you dont have backup from you main nas then don't do that kind of stuff, its way better to sink some money in hardware then learning all about lvm and mdadm data recovery to make things work again (external company for recovery is most often out of question because of pricing)

 

maybe a scenario with a raid1 with one internal and one external disk might be some carefree thing bat anything that goes above the used raid's spec's for loosing disks is very dangerous and not suggested

 

and to bridge to the answer from above, in theory usb and esata externel drives are handled the same, so it should be possible to configure esata ports in the same way as usb ports to work as internel ports, as esata is old technology and mostly in the way when it comes to  xpenology config files it's most often set to 0 and not in use, i used one esata port as internal port years back with dsm 6.x - but as a ootb solution is the thing then externel usb as internel drives is the common thing now and with 5 or 10 Gbits usb is just as capable as esata for a single disk (and you will have a good amount of usb ports on most systems where esata is usually, if there as any, is just one

 

if you want to use usb drives as "intrnal" drives then you can look here (arc loader wiki)

https://github.com/AuxXxilium/AuxXxilium/wiki/Arc:-Choose-a-Model-|-Platform

its listed specifically that its usable that way

 

 

[DSfuchs]

I can also confirm full USB functionality and advanced speed via TCRP. My systems are not only used for testing purposes. As of a year ago, ARPL didn't work.

 

It's also super easy and quick to migrate from internal RAID0 using a single Linux mdadm-command. If you have an external single drive, you should ideally place the parity disk of a RAID4 there. If you have a dual(or more)-case, it makes sense to place the mirror drives of the RAID10 (RAID0+1 with 2 near-copies) there. 

 

In general, the slow hard drive performance when writing in RAID 5/6 scenarios, with only 3-5 drives, is out of the question for me anyway. I can also imagine the long recovery times with the associated collapse in performance. We're not talking about a data center-sized installation where the penalty is almost negligible. Always remember that the higher the number of hard drives, the more likely (I love RAID4 there) it is that a hard drive will fail.

Edited April 26 by DSfuchs