1 |
Bill Kenworthy wrote: |
2 |
> On 2/1/20 10:27 am, Dale wrote: |
3 |
>> Mick wrote: |
4 |
>>> On Thursday, 2 January 2020 00:09:14 GMT Dale wrote: |
5 |
>>>> Howdy, |
6 |
>>>> |
7 |
>>>> As some may recall, I have a 8TB external SATA hard drive that I do |
8 |
>>>> back |
9 |
>>>> ups on. Usually, I back up once a day, more often if needed. |
10 |
>>>> Usually I |
11 |
>>>> turn the power on, mount it, do the back ups, unmount and turn the |
12 |
>>>> power |
13 |
>>>> back off. Usually it is powered up for 5 minutes or so. When I |
14 |
>>>> unmount |
15 |
>>>> it tho, I sometimes notice it is still doing something. I can feel |
16 |
>>>> the |
17 |
>>>> mechanism for the heads moving. It has a slight vibration to it. |
18 |
>>>> Questions are, what is it doing and should I let it finish before |
19 |
>>>> powering it off? I'd assume that once it in unmounted, the copy |
20 |
>>>> process |
21 |
>>>> is done so the files are safe. I guess it is doing some sort of |
22 |
>>>> internal checks or something but I'm not sure. |
23 |
>>> There is some delay with data still in the buffers between |
24 |
>>> rsync/cp/tar/what- |
25 |
>>> ever saying it's finished on your terminal and the drive itself |
26 |
>>> finishing |
27 |
>>> storing the data on the platters. |
28 |
>>> |
29 |
>>> If you look at vmstat, or keep an eye on Gkrelm you'll see what I mean. |
30 |
>>> Normally, if you try to unmount a drive while it is still being |
31 |
>>> written to, |
32 |
>>> the umount/udisks command will complain the drive is busy. |
33 |
>>> |
34 |
>> When it does it for a somewhat short period of time, I can understand |
35 |
>> that. It's one reason I try to leave it on when it "feels" that it is |
36 |
>> still busy. Thing is, there are times when it goes on for 30 minutes or |
37 |
>> more. At those times, even a USB stick should be done. One would think |
38 |
>> at least. It makes me curious as to what it is doing in that case. |
39 |
>> Still, I'd rather the unmount command force a wait until it is done. |
40 |
>> Honestly, I wouldn't want a drive or software that says something is |
41 |
>> done when it isn't. It's not good even when shutting a system down. |
42 |
>> Given the speed of drives, I would think a few seconds at most. Best to |
43 |
>> be safe. ;-) I just wonder, is it doing two different things? One |
44 |
>> when it is busy for short periods of time and something else when it |
45 |
>> goes on for a while. This is what sort of puzzles me. Selftest maybe?? |
46 |
>> |
47 |
>>>> Is it safe to turn it off even tho it is doing whatever it is doing? |
48 |
>>>> Should I wait? Does it matter? |
49 |
>>>> |
50 |
>>>> Thanks. |
51 |
>>>> |
52 |
>>>> Dale |
53 |
>>>> |
54 |
>>>> :-) :-) |
55 |
>>> If you wait for a few seconds after the backup is completed before |
56 |
>>> you unmount |
57 |
>>> the drive, you should be OK. Although it may slow down or any LEDs |
58 |
>>> flash less |
59 |
>>> frequently the drive may not stop spinning, unless there is some |
60 |
>>> power save |
61 |
>>> process taking control of it. |
62 |
>>> |
63 |
>> Given the speed, it is likely done when I tell the KDE thingy to |
64 |
>> unmount. Usually, I start the backup and walk away for a few minutes. |
65 |
>> I do it with one of my scripts, if one can call what I do a script, and |
66 |
>> it does the date command at the end. Even if there was a lot of |
67 |
>> changes, I can tell how long it was completed. I try to give it a |
68 |
>> couple minutes. Still, good point. This is one reason I'm asking about |
69 |
>> this. It's hard to know exactly what is going on here. |
70 |
>> |
71 |
>>>> P. S. Down to last router that was discussed in another thread so I |
72 |
>>>> bought it while they had it. Price may go up if I didn't. Did more |
73 |
>>>> research on old modem, it is risky to try to convert to AT&T. Some |
74 |
>>>> say |
75 |
>>>> not possible. |
76 |
>>> Right, ISP controlled firmware typically requires re-flashing the |
77 |
>>> device with |
78 |
>>> the new ISP's firmware version. In some cases even the boot code needs |
79 |
>>> replacing. Should you flash the router with a wrong firmware build, |
80 |
>>> you could |
81 |
>>> sometimes derive a door stop without additional cost. In this case |
82 |
>>> you'll |
83 |
>>> need a JTAG and access to its circuit board with an OEM |
84 |
>>> boot/firmware version |
85 |
>>> to recover it. In most cases OEMs support lines will redirect you |
86 |
>>> to your |
87 |
>>> ISP, who run an overseas support line and will ask you to reboot your |
88 |
>>> MSWindows PC ... O_o |
89 |
>>> |
90 |
>>> This is a reason I avoid these kind of routers as much as I can. |
91 |
>>> |
92 |
>> Keep in mind, two pieces of hardware. Router for the first two |
93 |
>> sentences and Modem for next two. Tried to be short so . . . . Anyway, |
94 |
>> router should be flashable with Openwrt. It's a slightly older model. |
95 |
>> New model may be ready for flashing in a year or two but not so much at |
96 |
>> the moment so I went with the older model. The modem, I never could find |
97 |
>> the firmware. I found links to it but those links ended up being dead. |
98 |
>> Even if I had it, it was unlikely to work. Possible but I'd be |
99 |
>> concerned about its stability and such even if it did take it. I have a |
100 |
>> modem and router on the way. I just didn't want to miss the deal on the |
101 |
>> router. They had several a couple weeks or so ago. I got the last |
102 |
>> one. Waiting for their arrival. |
103 |
>> |
104 |
>> Dale |
105 |
>> |
106 |
>> :-) :-) |
107 |
>> |
108 |
>> Oh, I may post and see if anyone needs a Frontier modem later. Maybe |
109 |
>> someone on here could use a spare or just needs one period, moving or |
110 |
>> something. Modem is wireless with a router as well. Nice modem I |
111 |
>> guess. |
112 |
>> |
113 |
> |
114 |
> Try atop from sys-process/atop - it will show you how busy individual |
115 |
> disks are (and a lot of other stats as well.) |
116 |
> |
117 |
> You can issue a sync command to flush any disk buffers before |
118 |
> unmounting (umounting should sync as well.). The heads may keep |
119 |
> moving because of the internal data management modern disks do. The |
120 |
> disks should be safe to power off despite this (they have an internal |
121 |
> flush/save/park routine on power loss, with enough energy stored to |
122 |
> take care of it) |
123 |
> |
124 |
> Bill |
125 |
|
126 |
Tried atop, it works best for this since it showed the drive device |
127 |
instead of process name like iotop. In my case, it shows sdj. I turned |
128 |
the drive on, atop showed nothing until I mounted it. I then did a |
129 |
quick back up and it showed up like it should. Once done, it showed no |
130 |
activity until I told KDE to unmount. After it was unmounted, it showed |
131 |
no further activity according to atop, iotop or anything else for that |
132 |
matter. I could however feel it doing something. The heads were moving |
133 |
around every few seconds or so. Sometimes close together, sometimes |
134 |
several seconds or more apart. It's sort of intermittent type activity. |
135 |
|
136 |
I might add, the light that shows activity on the enclosure shows |
137 |
nothing either. The light is odd, it's always on but blinks when there |
138 |
is activity. Usually they are off and come on when active. Sort of |
139 |
weird but OK. |
140 |
|
141 |
I think once unmounted, the files are done, or a short time afterwards. |
142 |
After that, I think you and other are right, it's doing some internal |
143 |
stuff. |
144 |
|
145 |
Thanks again to all who replied. I'm not 100% sure of what it is doing |
146 |
but pretty sure I can power it off once the back up is done and it is |
147 |
unmounted. |
148 |
|
149 |
Dale |
150 |
|
151 |
:-) :-) |