1 |
Rich Freeman wrote: |
2 |
> On Mon, Aug 3, 2020 at 8:24 AM Dale <rdalek1967@×××××.com> wrote: |
3 |
>> In the past, I've never seen the drive on the larger files be that slow even toward the end. Generally, it stays pretty close to 180MBs/sec or so which is what I usually get with PMR drives. |
4 |
> Yeah, just hard to be certain without ditching the filesystem layer or |
5 |
> doing some kind of comparison. The difference in write speed across |
6 |
> the drive on recent drives I've gotten is more pronounced than I've |
7 |
> seen on other drives, but these drives tend to be around 12TB. |
8 |
> |
9 |
> Definitely the thing to watch out for is a big drop in transfer rate |
10 |
> once a large number of blocks have been transferred continuously, and |
11 |
> then performance returns after you let the drive thrash for a while. |
12 |
> I've seen complaints of zfs rebuilds going from hours/days to |
13 |
> weeks/months in length, so it isn't just a 50% drop when you're doing |
14 |
> worst-case access patterns. On the other hand I hear that mdadm isn't |
15 |
> so bad, so if the writes are sequential the drive might be better at |
16 |
> skipping the cache, and maybe zfs just does its rebuild |
17 |
> non-sequentially (which isn't really ideal anyway). |
18 |
> |
19 |
> I haven't really dug into the guts of how zfs metadata works, but with |
20 |
> btrfs I believe the chunks are basically their own layer, and the |
21 |
> filesystem can scrub them without really any care about what files are |
22 |
> stored in them. That allows them to be easily scrubbed sequentially. |
23 |
> When I did rebuilds on btrfs they tended to run at about the max |
24 |
> throughput of the drives as long as there wasn't any other access |
25 |
> going on. It can also do read-only scrubs to check data integrity |
26 |
> sequentially across the disk, which suggests the checksums are stored |
27 |
> at a lower layer and so the data can be verified without worrying |
28 |
> about file fragmentation and so on. This layering also lets btrfs |
29 |
> switch "RAID modes" on the fly with half of the disk being RAID1 and |
30 |
> half the disk being RAID5 and so on - each region of a disk is |
31 |
> independent from the others and so mode changes only impact new |
32 |
> regions until you do a full rebalance to rewrite everything. Of |
33 |
> course, zfs has its own advantages. |
34 |
> |
35 |
|
36 |
This drive is formatted with ext4. It doesn't have LVM or anything just |
37 |
straight ext4. Given it is external, I didn't see the point of having |
38 |
LVM on it and adding another layer to deal with when there is no |
39 |
benefits to it. While it does delete some files and overwrite others, |
40 |
it mostly adds new files. I'd guess it just throws them on the end but |
41 |
who knows what it is really doing under the hood that neither of us |
42 |
knows about. ;-) |
43 |
|
44 |
I did notice that is has reached 80% full. I'm going to start figuring |
45 |
out what not to backup and such pretty soon. If I have my videos, I'm |
46 |
good. I could backup Documents and some other OS related stuff on |
47 |
another drive. I usually backup /etc and the world file. Hmmm, may |
48 |
want to grab my local overlay too. I hadn't thought about that. Funny |
49 |
how typing in a email makes me think of things like that. |
50 |
|
51 |
Maybe when I do a large backup next time, I'll think to save the output |
52 |
so I can share it. That might shed some light on the situation. I just |
53 |
thought it was interesting that when it hit about 50 or 60GBs of data it |
54 |
got a good deal slower and then did the bumpy thing a lot longer too. |
55 |
To me, I figured it ran out of PMR space and was slinging stuff good |
56 |
trying to catch up. There was several files where it just sat there, |
57 |
for many seconds with nothing moving. Usually, 10 seconds is about the |
58 |
longest wait but it is a large file, movie or something that is GBs or |
59 |
so. I'd guess close to a minute for some this last time. Big |
60 |
difference. Most files were around 300MBs too. Found a source for good |
61 |
HD stuff that isn't to large. ;-) |
62 |
|
63 |
Anyway, thought it worth mentioning. The drive serves its purpose well |
64 |
enough for what it does. Still wouldn't want it somewhere more critical |
65 |
tho. Sticking with PMR/CMR until they get things sorted out better. If |
66 |
I bought another drive just for external backup use tho, I might get a |
67 |
SMR if the price was right. It wouldn't be my first choice tho. |
68 |
|
69 |
Dale |
70 |
|
71 |
:-) :-) |