1 | Mondo for Debian
|
---|
2 | ================
|
---|
3 |
|
---|
4 | This text contains hints, workarounds and explanations for using mondo on a
|
---|
5 | Debian system.
|
---|
6 |
|
---|
7 | Location of files mondorescue.iso, MA.log.gz & changed.files
|
---|
8 | ------------------------------------------------------------
|
---|
9 |
|
---|
10 | The above files are all located in /var/cache/mindi irrespective of what
|
---|
11 | the upstream documentation may say.
|
---|
12 |
|
---|
13 |
|
---|
14 | Location of file mondorestore.log
|
---|
15 | ---------------------------------
|
---|
16 |
|
---|
17 | The above file is located in /var/log irrespective of what the upstream
|
---|
18 | documentation may say.
|
---|
19 |
|
---|
20 |
|
---|
21 | Kernel Panic when booting restore media with 2.6.18 kernel
|
---|
22 | ----------------------------------------------------------
|
---|
23 |
|
---|
24 | There may be a kernel panic when booting a restore media containing a
|
---|
25 | stock 2.8.18 Debian kernel.
|
---|
26 |
|
---|
27 | The workaround is to remove 'acpi=off' from the 'ADDITIONAL_BOOT_PARAMS'
|
---|
28 | variable defined at the beginning of /usr/sbin/mindi.
|
---|
29 |
|
---|
30 | This may only affect users with certain VIA chipsets. Please see Debian
|
---|
31 | bug #389931 for more details.
|
---|
32 |
|
---|
33 | Note: You will have to boot into the rescue media to verify you are not
|
---|
34 | affected by this. QEMU or VMware do not expose the problem!
|
---|
35 |
|
---|
36 |
|
---|
37 | NTFS Windows Installations not bootable after restore
|
---|
38 | -----------------------------------------------------
|
---|
39 |
|
---|
40 | After a restore, NTFS Windows installations may not be bootable. Rather an
|
---|
41 | error message "A disk read error occurred. Press Xtrl+Alt+Del to restart." is
|
---|
42 | displayed.
|
---|
43 |
|
---|
44 | A fix for this is to use gparted (and possibly gtparted but no other
|
---|
45 | partitioning tool) resize the NTFS filesystem to that it fills the entire
|
---|
46 | partition it lives in. The error appears to only happen for NTFS filesystems
|
---|
47 | that are smaller than the partition they live in.
|
---|
48 |
|
---|
49 |
|
---|
50 | Boot Images in /root/images
|
---|
51 | ---------------------------
|
---|
52 |
|
---|
53 | MondoRescue does not currently allow for influencing where the boot images it
|
---|
54 | creates are stored. They will always end up under /root/images. This can be a
|
---|
55 | problem on installations where the partition that holds /root is small.
|
---|
56 |
|
---|
57 | A workaround is to create a symbolic link to a different location with more
|
---|
58 | space. Example:
|
---|
59 |
|
---|
60 | /root/images -> /home/mondorescue/images
|
---|
61 |
|
---|
62 |
|
---|
63 | CD Streaming
|
---|
64 | ------------
|
---|
65 |
|
---|
66 | The CD streaming feature of mondoarchive (option '-C') is EXPERIMENTAL which
|
---|
67 | is what the manpage states. In fact, it appears to be rather broken. Some of
|
---|
68 | the issues are fixed in this package: selecting the burn speed should work,
|
---|
69 | it won't segfault during a mondoarchvie run. However, restoring from multiple
|
---|
70 | CDs still doesn't seem to work but hangs towards the end of the first CD.
|
---|
71 |
|
---|
72 | If you want to experiment with this feature, it might help to keep the
|
---|
73 | following in mind:
|
---|
74 |
|
---|
75 | - Use blank CDs only. If you use CDRWs (which is recommended for
|
---|
76 | environmentally and monetarian reasons by the author of this text) make sure
|
---|
77 | they are blank when throwing them at mondoarchive in streaming mode.
|
---|
78 | - Don't make mistakes when feeding media to mondoarchive in streaming mode. It
|
---|
79 | will choke and the backup will have to be restarted.
|
---|
80 |
|
---|
81 | Having said that, streaming is certainly a useful feature. I'll work on it as
|
---|
82 | time permits. If your time does permit, I am receptive to patches (and so is
|
---|
83 | upstream).
|
---|
84 |
|
---|
85 |
|
---|
86 | Image Archiving: (V)FAT versus NTFS
|
---|
87 | -----------------------------------
|
---|
88 |
|
---|
89 | If you have a (V)FAT partition please mount it to get it backed up by
|
---|
90 | mondoarchive and restored by mondo. Using the '-x' option will not work.
|
---|
91 |
|
---|
92 | For NTFS, '-x' is the way to go.
|
---|
93 |
|
---|
94 |
|
---|
95 | "Too many open files" in log
|
---|
96 | ----------------------------
|
---|
97 |
|
---|
98 | You may get entries in the log similar to this one:
|
---|
99 |
|
---|
100 | getfacl: /home/andree/shed/kernel/kernel-source-2.4.26/include/config/paride/pf: Too many open files
|
---|
101 |
|
---|
102 | If you do, increase the number of allowed open file descriptors for the user
|
---|
103 | session. The default value for open file descriptors in Debian is 1024. To
|
---|
104 | increase this value, if you run bash (and potentially other shells), use
|
---|
105 | 'ulimit -n 2048' to double this.
|
---|
106 |
|
---|
107 |
|
---|
108 | Using the File-Browser for Partial Restore
|
---|
109 | ------------------------------------------
|
---|
110 |
|
---|
111 | When using mondorestore to selectively restore files, the 'Editing filelist'
|
---|
112 | window will only list a limited, hardcoded number of files. This number is
|
---|
113 | defined by define 'ARBITRARY_MAXIMUM' in include 'my-stuff.h'. This value is
|
---|
114 | currently set to 500 in the upstream source and has been increased to 2000 for
|
---|
115 | the Debian package. Changing this value requires a recompile of the package.
|
---|
116 |
|
---|
117 | A workaround for this shortcoming is to restore a complete directory tree to a
|
---|
118 | temporary location, delete all unwanted files and directories, and move the
|
---|
119 | remainder to its final destination.
|
---|
120 |
|
---|
121 |
|
---|
122 | Using floppy disks as rescue media
|
---|
123 | ----------------------------------
|
---|
124 |
|
---|
125 | mondoarchive will offer to write a set of rescue floppies (unless the '-F'
|
---|
126 | switch is used). These floppy disks will only work with the FAILSAFE kernel. The
|
---|
127 | reason is that stock Debian kernels don't have the floppy driver built into the
|
---|
128 | kernel and that the kernel is too large for one floppy with an initrd and a
|
---|
129 | kernel image. Because of the space issue, mondoarchiv (or rather mindi) creates
|
---|
130 | a root/boot floppy set. Because the stock kernel does not have floppy support
|
---|
131 | compiled in a kernel panic results when trying to acces the second (the root)
|
---|
132 | floppy.
|
---|
133 |
|
---|
134 | Also, building the boot/root floppy images requires that lilo is installed on
|
---|
135 | the system. This does not mean that it actually needs to be used to boot the
|
---|
136 | system, only that the package is installed. (Because most people won't need the
|
---|
137 | boot/root floppies, the mindi package only suggests the lilo package.)
|
---|
138 |
|
---|
139 | Finally a word of caution: Floppy disks are notoriously unreliable. Dont't use
|
---|
140 | them unless you really really have to. (A CD writer costs the same as a few
|
---|
141 | boxes of floppy disks these days.)
|
---|
142 |
|
---|
143 | -- Andree Leidenfrost <andree@debian.org> Sat, 15 Sep 2007 21:45:37 +1000
|
---|