Ticket #628 (assigned defect)
3.0.2-1 Fail to Restore from External Hard Disk
| Reported by: | SamK | Owned by: | bruno |
|---|---|---|---|
| Priority: | high | Milestone: | 3.0.4 |
| Component: | mondo | Version: | 3.0.2 |
| Severity: | major | Keywords: | |
| Cc: | bs27975.mondorescuetrac@… |
Description
This ticket is raised at the request of Bruno as a result of a mailing list discussion. See archive: Date=June 10 2012 Title=3.0.2-1 Fail to Restore from External Hard Disk
A single mondorescue-1.iso backup is successfully created directly to an external hard disk and stored in /mondo.
Booting from the mondorescue.iso (mindi) CDROM
- Boot codes: interactive noresize
- Read from=Hard disk
- Prefix=mondorescue
- On what device do the ISO files live=/dev/sdb1
- What is disk format=ext4
- Are you restoring from an external hard drive=Yes
- At what path on this device can the ISO files be found=/mondo
- When mondorestore displays the message "Failed to find config/file archives. Choose another source?"
- ALT-F2
- mount "... /dev/sdb1 on /tmp/isodir type ext4 (ro,relatime,barrier=1,data=ordered)"
- ls -1 /mnt/isodir/mondo/*.iso "/tmp/isodir/mondo/mondorescue-1.iso /tmp/isodir/mondo/mondorescue.iso"
It seems that mondorestore mounts the external disk OK but does not use the path to the directory holding the backup ISO.
At the suggestion of Bruno mondorescue-1.iso was copied to the root of the external drive and the restore attempted again. This time the ISO is found but the restore fails at a later stage.
Booting from the mondorescue.iso (mindi) CDROM
- Boot codes: interactive noresize
- Read from=Hard disk
- Prefix=mondorescue
- On what device do the ISO files live=/dev/sdb1
- What is disk format=ext4
- Are you restoring from an external hard drive=Yes
- At what path on this device can the ISO files be found=/
- Editing Mountlist=OK
- Save mountlist and continue=Yes
- Erase and partition hard drives=Yes
- Restore all data=Yes
- A message is displayed "Please insert ISO #1 and press Enter"
At this point the mondorestore.log was saved as it is not possible to proceed further. The log is attached.
Attachments
Change History
comment:1 Changed 11 months ago by bs27975
Ditto, here.
First: Having followed FAQ 42 and written the mondoarchive.iso to a usb key, when booted, fdisk was not present.
- writing the iso out to a cdrom instead and booting from it, all was 'well.' (fdisk found, etc.)
So, the built usb did not build / boot correctly (syslinux booted but busybox didn't change into the loaded files)?
- curiously, booting from usb, it could not find the cdrom (which would be empty at the time).
When booting from the mondoarchive created mondorescue.iso:
- from what I could see, despite finding the .iso's on /dev/sdb6 [as evidenced by being listed at /tmp/isos (path?)], mondorescue ejected the (boot) cd, then asked for the first cd iso to be inserted.
- 'mount /path/to/mondorescue-1.iso /cdrom -o loop' got me beyond the request for the first iso, but then I went into an endless loop of it trying to build grub. I got the sense that it would drop me to a command prompt so I could manually repair things, but the screen merely blinked then took me to the next prompt. Since I had not fixed anything (no opportunity, no command line), I was in an endless loop [fix grub, (no) command line, fix grub, <repeat>], I had to reboot.
- post attempt 3 - I now understand this to have been the post-iso restore sequence. It did not read iso 2, and skipped to the post-iso restore stage.
Attempt number 2, specifying 'interactive noeject text':
- I got to a Yes / No prompt, and none of the following were accepted: Yes, Y, y, yes. This was somewhere around the mount list questions.
- rebooted, as nothing was being accepted, preventing me from proceeding.
- Again, post attempt 3, I now realize this to have been the post-iso restore process. However, the notext mode would not accept yes, and I bailed.
Attempt number 3, 'interactive noeject' with no reparitioning or formatting (as they were done in attempt 1), same situation (leading to this bug report - request for insertion of disc, when iso's are on an external usb disk)
- 'mount /path/to/mondorescue-1.iso /mnt/cdrom -o loop' (Alt-F2) used this time, per 'tail mondoarchive.log' (Difference from attempt 1: /mnt/cdrom instead of /cdrom. I believed at the time of attempt 1 that /cdrom was correct - I believe I saw a /cdrom directory, and so assumed that directory should be used for mounting, at that time. Point being, here: It appears being unable to find iso 1, it gave up and went on to post-iso restore activities.)
- repeat for each additional iso. /mnt/cdrom being automatically unmounted. i.e. no umount /mnt/cdrom necessary, it already having been done.
- restore completed successfully. (With 1 read error.)
Changed 11 months ago by bs27975
-
attachment
BS1-mondorestore1.log
added
T628-BS mondo restore log - attempt 1 (failed)
Changed 11 months ago by bs27975
-
attachment
BS3mondorestore3.7z
added
T628-BS mondo restore log - attempt 3 (kludge successful)
comment:4 Changed 7 weeks ago by bruno
From another user:
mondorestore *at the final stage* attempts to restore from a file called mondorescue-1.iso, although we have explicitly stated that the backup prefix is mondonetvis (so it should be looking for mondonetvis-1.iso, as it has correctly done at all previous stages of the process).
To prove the case, I changed the name of the backup file to mondorescue-1.iso and this time the restore process was successful. So there is a little bug there: mondorestore does not use the custom backup name at the final stage of the restore process, but the default one ("mondorescue").
comment:5 follow-up: ↓ 6 Changed 7 weeks ago by bruno
However he mentioned this doesn't seem to be related directly. However worth looking at.
comment:6 in reply to: ↑ 5 Changed 7 weeks ago by SamK
Replying to bruno:
However he mentioned this doesn't seem to be related directly. However worth looking at.
Re-tested using names: mondorescue mondorescueverass-single The results are similar to those obtained in the original bug report, i.e. the problem is still present as reported.
