encrypted lvm initialisation is very slow

Bug #151244 reported by Ian Jackson on 2007-10-10
2
Affects Status Importance Assigned to Milestone
partman-crypto (Ubuntu)
Medium
Colin Watson
Hardy
Medium
Colin Watson

Bug Description

Binary package hint: debian-installer

I'm doing a test install of gutsy d-i i386 20071009.1 on a few-years-old laptop with a 40G hard disk.

I have selected "use whole disk with LVM and encryption". Currently the system is reporting that it is "Erasing data on SCSI1 (0,0,0), partition #5 (sda)". It's about 44% complete after about an hour.

This laptop normally has adequate performance for installation and use. Is it really necessary for the install to read/write data to the whole disk ?

Henrik Nilsen Omma (henrik) wrote :

This has also been remarked by others we are looking at turning off the disk-wipe post-RC.

Changed in partman-crypto:
importance: Undecided → Medium
status: New → Confirmed
Martin Pitt (pitti) wrote :

This was discussed yesterday, and Colin kindly agreed to work on this today and get the default erase flag flipped to off.

(Ab)using the -rc milestone for this to keep it on the radar.

Changed in partman-crypto:
assignee: nobody → kamion
milestone: none → ubuntu-7.10-rc
Colin Watson (cjwatson) on 2007-10-12
Changed in partman-crypto:
milestone: ubuntu-7.10-rc → ubuntu-7.10
Colin Watson (cjwatson) wrote :

partman-crypto (21ubuntu3) gutsy; urgency=low

  * Disable partition erasing by default, as it's very slow and only of
    moderate value (LP: #151244).
  * Warn that the passphrase cannot be recovered if lost (helps with LP
    #151264).

 -- Colin Watson <email address hidden> Fri, 12 Oct 2007 14:51:34 +0100

Changed in partman-crypto:
status: Confirmed → Fix Released
Martin Pitt (pitti) wrote :

Reopening. It's back on Alpha-2 (merge regression?)

Changed in partman-crypto:
milestone: ubuntu-7.10 → none
status: Fix Released → Confirmed
status: Confirmed → Triaged
Colin Watson (cjwatson) wrote :

This is intentional; the erase step is cancellable now (or should be; if it isn't, *then* please reopen this ...), which I think is a more elegant solution to the original problem.

I did document this in the changelog. :-)

Changed in partman-crypto:
status: Triaged → Fix Released
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers