1 |
2013/9/3 <meino.cramer@×××.de> |
2 |
|
3 |
> Francisco Ares <frares@×××××.com> [13-09-04 02:08]: |
4 |
> > Em 03/09/2013 13:12, <meino.cramer@×××.de> escreveu: |
5 |
> > > |
6 |
> > > William Kenworthy <billk@×××××××××.au> [13-09-03 17:16]: |
7 |
> > > > On 03/09/13 11:26, meino.cramer@×××.de wrote: |
8 |
> > > > > William Kenworthy <billk@×××××××××.au> [13-09-03 05:08]: |
9 |
> > > > >> On 03/09/13 10:45, meino.cramer@×××.de wrote: |
10 |
> > > > >>> walt <w41ter@×××××.com> [13-09-03 04:15]: |
11 |
> > > > >>>> On 09/02/2013 09:15 AM, meino.cramer@×××.de wrote: |
12 |
> > > > >>>>> The rootfs and $HOME of my embedded system is stored |
13 |
> > > > >>>>> on a 16GB SD-card (about 5GB used, rest free). The FS |
14 |
> > > > >>>>> is ext4. |
15 |
> > > > >>>>> |
16 |
> > > > >>>>> Since the system hangs for unknown reasons several times |
17 |
> > > > >>>> Does it hang at a predictable point, like during boot, or |
18 |
> poweroff? |
19 |
> > > > >>>> |
20 |
> > > > >>>> I know almost nothing about SD cards (yet). Do they develop bad |
21 |
> > > > >>>> blocks like other storage media? I notice fsck.ext4 has a -c |
22 |
> flag |
23 |
> > > > >>>> to check for bad blocks. |
24 |
> > > > >>>> |
25 |
> > > > >>> No, it hangs while compiling or while updateing (eix-sync; emerge |
26 |
> > ...). |
27 |
> > > > >>> |
28 |
> > > > >>> |
29 |
> > > > >>> I did the following now: |
30 |
> > > > >>> I did a binary image backup with dd of the sdcard. |
31 |
> > > > >>> I made a backup of the all files from the bad fs with tar. |
32 |
> > > > >>> I say "YES" to fsck to fix what it found. |
33 |
> > > > >>> I made another backup of the all files from the bad fs with tar. |
34 |
> > > > >>> I md5summed both tar archives and found them identical. |
35 |
> > > > >>> |
36 |
> > > > >>> Now...is the conclusion correct, that the identical md5sum |
37 |
> > > > >>> indicate, that the fixed error of the fs only had impact to |
38 |
> > > > >>> already invalidated data? |
39 |
> > > > >>> Or whatelse could this indicate? |
40 |
> > > > >>> |
41 |
> > > > >>> Best regards, |
42 |
> > > > >>> mcc |
43 |
> > > > >>> |
44 |
> > > > >>> PS: What come mind just in this moment: |
45 |
> > > > >>> Can I ran fsck on an binary image of the fs which I made with dd |
46 |
> > somehow? |
47 |
> > > > >>> |
48 |
> > > > >>> |
49 |
> > > > >>> |
50 |
> > > > >>> |
51 |
> > > > >>> |
52 |
> > > > >> Have you run out of inodes? - ext 4 has had very mixed success for |
53 |
> > me on |
54 |
> > > > >> solid state. Running out of inodes is a real problem for gentoo |
55 |
> on |
56 |
> > > > >> smaller SD cards with standard settings. |
57 |
> > > > >> |
58 |
> > > > >> BillK |
59 |
> > > > >> |
60 |
> > > > >> |
61 |
> > > > >> |
62 |
> > > > > Does this error message from fsck indicate that? I am really bad in |
63 |
> > > > > guessing what fsck tries to cry at me ... ;) |
64 |
> > > > > |
65 |
> > > > > |
66 |
> > > > >>> solfire:/root>fsck.ext4 -f -p /dev/sdb2 |
67 |
> > > > >>> rootfs: Inodes that were part of a corrupted orphan linked |
68 |
> list |
69 |
> > found. |
70 |
> > > > >>> |
71 |
> > > > >>> rootfs: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY. |
72 |
> > > > >>> (i.e., without -a or -p options) |
73 |
> > > > >>> [1] 18644 exit 4 fsck.ext4 -f -p /dev/sdb2 |
74 |
> > > > >>> |
75 |
> > > > >>> |
76 |
> > > > > Is there any way to correct the settings from the default values to |
77 |
> > > > > more advances ones, which respect the sdcard size of 16GB *without* |
78 |
> > > > > blanking it...a "correction on the fly" so to say??? |
79 |
> > > > > |
80 |
> > > > > And if not: Is there a way to backup the sdcard and playback the |
81 |
> files |
82 |
> > > > > after reformatting it by preserving all three time stamps of the |
83 |
> > > > > files (atime is deactivated via fstab though) ? |
84 |
> > > > > |
85 |
> > > > > Best regards, |
86 |
> > > > > mcc |
87 |
> > > > > |
88 |
> > > > > |
89 |
> > > > > |
90 |
> > > > > |
91 |
> > > > > |
92 |
> > > > df -i - if you get 100% iUSE or near to it thats your problem ... I |
93 |
> have |
94 |
> > > > seen that error message you give as a result of running out of inodes |
95 |
> > > > corrupting the FS. |
96 |
> > > > |
97 |
> > > > No, your only way out is to copy (I use rync) the files off, recreate |
98 |
> > > > the fs with max inodes ("man mke2fs") and rsync the files back. |
99 |
> Once an |
100 |
> > > > ext* fs has been created with a certain number of inodes its fixed |
101 |
> until |
102 |
> > > > you re-format. |
103 |
> > > > |
104 |
> > > > I get it happening regularly on 4G cards when I forget and just |
105 |
> emerge a |
106 |
> > > > couple of packages without cleaning up in between packages. On 16G |
107 |
> > > > cards, its compiling something like glibc or gcc that uses huge |
108 |
> numbers |
109 |
> > > > of inodes at times. On a single 32G card I have, the standard |
110 |
> settings |
111 |
> > > > have been fine ... so far :) |
112 |
> > > > |
113 |
> > > > Billk |
114 |
> > > > |
115 |
> > > > |
116 |
> > > |
117 |
> > > df -i gives the following: |
118 |
> > > |
119 |
> > > rootfs 971040 352208 618832 37% / |
120 |
> > > /dev/root 971040 352208 618832 37% / |
121 |
> > > devtmpfs 63420 434 62986 1% /dev |
122 |
> > > tmpfs 63456 389 63067 1% /run |
123 |
> > > shm 63456 1 63455 1% /dev/shm |
124 |
> > > cgroup_root 63456 6 63450 1% /sys/fs/cgroup |
125 |
> > > /dev/mmcblk0p1 0 0 0 - /boot |
126 |
> > > |
127 |
> > > |
128 |
> > > You mentioned rsync to backup... |
129 |
> > > |
130 |
> > > I used |
131 |
> > > |
132 |
> > > sudo tar cvf <backup file> <root of embedded system> |
133 |
> > > |
134 |
> > > the rootfs has only one partition... |
135 |
> > > |
136 |
> > > Is it alos ok to use tar or is there any drawback....? |
137 |
> > > |
138 |
> > > Best regards, |
139 |
> > > mcc |
140 |
> > > |
141 |
> > > |
142 |
> > > |
143 |
> > |
144 |
> > There are some parameters for creating a better backup archive using tar, |
145 |
> > like --same-owner and --atime- preserve. |
146 |
> > |
147 |
> > By the way, it would be an interesting project to export some folders on |
148 |
> > your home computer using nfs, tuneling it through ssh, monting it locally |
149 |
> > in your embedded computer, and applying an unionfs to the rootfs. Just |
150 |
> > dreaming, of course. |
151 |
> > |
152 |
> > Góod luck |
153 |
> > Francisco |
154 |
> |
155 |
> Hi Francisco, |
156 |
> |
157 |
> as I understand the man page, --same-owner is only activ while |
158 |
> extracting a tar: |
159 |
> |
160 |
> --same-owner |
161 |
> create extracted files with the same ownership |
162 |
> |
163 |
> while extracting I always use |
164 |
> |
165 |
> --preserve |
166 |
> like --preserve-permissions plus --same-order |
167 |
> |
168 |
> . Atime setting is disabled via fstab on my embedded system for two |
169 |
> reasons: |
170 |
> Performance wise since any access to a file will trigger a write |
171 |
> action to the flash chip even when reading the file. |
172 |
> Any write action to a flash chip wear out the chip -- it has a limited |
173 |
> number of write cycles. |
174 |
> I also disbaled atime on my PC for the first reason. |
175 |
> |
176 |
> What makes the unionfs'ed nfs mount of my PC on the embedded system |
177 |
> interesting to you ? |
178 |
> (sorry if this question sounds bad/negative/... or so...its my limited |
179 |
> english. Its simply and only a question and the wish of getting more |
180 |
> infos... :) |
181 |
> |
182 |
> Best regards, |
183 |
> mcc |
184 |
> |
185 |
> |
186 |
> |
187 |
> |
188 |
> |
189 |
> |
190 |
Hi, Meino. |
191 |
|
192 |
Sorry for my delay in answering your message. Sorry for my english, too, as |
193 |
a non-native speaker, I know sometimes I may sound strange. |
194 |
|
195 |
I am (trying to) finishing an embedded equipment, using an Intel x64, and |
196 |
Gentoo Linux. The main file system will be stored in a SATA "disk on |
197 |
module" flash device, but there are some directories that are not needed |
198 |
for daily use, or should not be present at all on the final product, like |
199 |
private source code used to build the program that run this equipment. |
200 |
|
201 |
So on the development system, I have used several disk partitions (as a |
202 |
first approach) for this directories, like /usr/portage , /usr/src , |
203 |
/usr/include and so on, and I was thinking on a way of remote access, so |
204 |
that a remote system could use the structure of this local development |
205 |
system. So I suppose that some unionfs mounts would make things appear |
206 |
local to the remote system. But probably just nfs would do the trick. As I |
207 |
said, just dreaming. |
208 |
|
209 |
And what about your problem? |
210 |
|
211 |
Best regards, |
212 |
Francisco |