1 |
On Mon, Mar 2, 2020 at 12:22 AM William Kenworthy <billk@×××××××××.au> wrote: |
2 |
> |
3 |
> I thought lizardfs was much more community minded |
4 |
> but you are characterising it as similar to moosefs - a taster offering |
5 |
> by a commercial company holding back some of the non-essential but |
6 |
> jucier features for the paid version - is that how you see them? |
7 |
|
8 |
I don't see much of an active community. It seems like most actual |
9 |
development happens outside of the public repo, with big code drops by |
10 |
the private team doing the work (which seems to be associated with a |
11 |
company). A bit like the Android model. I'm sure they'll accept pull |
12 |
requests, but that isn't how most of the work is getting done. |
13 |
|
14 |
It seems like the main difference between them and moosefs is that |
15 |
they're making more stuff FOSS to entice users over. Shadow masters |
16 |
are FOSS as opposed to just having metadata loggers. HA is FOSS in |
17 |
the latest RC. |
18 |
|
19 |
So, it seems like their model is to trickle out the non-free stuff and |
20 |
make it free after a delay. |
21 |
|
22 |
It really seems like Ceph is the best fully open platform out there, |
23 |
but the resource requirements just make it impractical. I have no |
24 |
doubt that it can scale FAR better with its design, but that design |
25 |
basically forces every node to be a bit of a powerhouse, versus |
26 |
Lizardfs where you just have one daemon with all the intelligence and |
27 |
the rest are just dumping files on disks. And you really don't need |
28 |
much CPU/RAM for the master if you're serving large files - the |
29 |
demands would go up with IOPS and number of files, and multimedia is |
30 |
low on both. |
31 |
|
32 |
> By the way, to keep to the rpi subject, I did have a rpi3B with a usb2 |
33 |
> sata drive attached but it was hopeless as a chunkserver impacting the |
34 |
> whole cluster. Having the usb data flow and network data flow through |
35 |
> the same hub just didn't go well |
36 |
|
37 |
Hard drives plus 100Mbps LAN sharing a single USB 2.0 hub is |
38 |
definitely not a recipe for NAS success... |
39 |
|
40 |
When I upgraded to UniFi switches I really only noticed for the first |
41 |
time how many hosts I have that aren't gigabit, and they're mostly Pis |
42 |
at this point. They're nice little project boards but for anything |
43 |
IO-intensive they're almost always the wrong choice. |
44 |
|
45 |
The RockPro64 I'm using has gigabit plus PCIe 3.0 x8 plus USB3 and as |
46 |
far as I can tell they don't have any contention. Maybe they're all |
47 |
on a PCIe bus or something but obviously that can handle quite a bit. |
48 |
Only issue was that the rk3399 PCIe drivers were not the most robust |
49 |
in the kernel, but ayufan and the IRC channel were both helpful and |
50 |
his kernel branch is actively maintained, so I was able to get |
51 |
everything sorted (some delays needed during training to allow boards |
52 |
to initialize and I was having power issues in the beginning). Much |
53 |
of the rk3399 support in the kernel was pushed by Google for |
54 |
Chromebooks and LSI HBAs weren't exactly on their list of things to |
55 |
test with those - not sure if the Chromebooks put much of anything on |
56 |
PCIe. |
57 |
|
58 |
-- |
59 |
Rich |