Merge lp:~canonical-platform-qa/ubuntu-test-cases/server-preseed into lp:ubuntu-test-cases/server
| Status: | Merged |
|---|---|
| Merged at revision: | 103 |
| Proposed branch: | lp:~canonical-platform-qa/ubuntu-test-cases/server-preseed |
| Merge into: | lp:ubuntu-test-cases/server |
| Diff against target: |
237 lines (+64/-0) 16 files modified
preseeds/default.preseed (+4/-0) preseeds/dns-server.preseed (+4/-0) preseeds/lamp-reboot.preseed (+4/-0) preseeds/lamp.preseed (+4/-0) preseeds/lvm.preseed (+4/-0) preseeds/mail-server.preseed (+4/-0) preseeds/minimal-virtual.preseed (+4/-0) preseeds/multi-lvm.preseed (+4/-0) preseeds/openssh-server.preseed (+4/-0) preseeds/postgresql-server.preseed (+4/-0) preseeds/print-server.preseed (+4/-0) preseeds/raid1.preseed (+4/-0) preseeds/samba-server.preseed (+4/-0) preseeds/templates/base (+4/-0) preseeds/tomcat-server.preseed (+4/-0) preseeds/virtual-host.preseed (+4/-0) |
| To merge this branch: | bzr merge lp:~canonical-platform-qa/ubuntu-test-cases/server-preseed |
| Related bugs: |
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Scott Moser | 2016-01-12 | Approve on 2016-01-19 | |
| Jean-Baptiste Lallement (community) | Approve on 2016-01-12 | ||
|
Review via email:
|
|||
Description of the Change
This adds unmount active to the preseeds to work around recent install troubles.
| Scott Moser (smoser) wrote : | # |
Why would there be mounted partitions during an install?
Can you give a pointer to what was failing?
I'm just trying to understand what changed that caused this.
| Scott Moser (smoser) wrote : | # |
further searching shows likely that the bug is http://
The fix here would indicate that the other google hit for 'unmount_active' is fixed (http://
Also, any reason to not '--fixes lp:1533243' on this MP?
| Scott Moser (smoser) wrote : | # |
Well, if this actually *does* fix the problem, then I think it is fine to pull it until bug 1533243 is fixed. I'd like a 'FIXME: remove after bug 1533243 fixed' style comment in the preseeds.
And then also, if it does fix the problem, then we should close bug 1355845 as fix-released.
| Max Brustkern (nuclearbob) wrote : | # |
I think bug 1355845 is fixed now. I'm not sure about bug 1533243. The same preseed worked a few months ago, and has been working for years. If someone wants to say, "yes, this is how it works now," then I think we can mark bug 1533243 as won't fix, and use the new preseed as the solution to the problem. Either way, I don't think updating the test preseeds actually fixes 1533243, it just fixes a utah bug I hadn't filed yet because we're still working through a couple other preseed problems there.
- 104. By Max Brustkern on 2016-01-19
-
Added workaround note
| Scott Moser (smoser) wrote : | # |
So... My belief at this point is that in fact partman/
a.) is not preseedable
b.) does not do what we thought it did
My current understanding is that the tests were failing after installation was done and when d-i was trying to unmount filesystems. Because of a backgrounded task spawned by 'apt-get update'(described in bug 1527710) that unmount would fail.
The changes here actually did not do anything, and evidence shows that.
We continued to see the same errors after this change was merged.
| Max Brustkern (nuclearbob) wrote : | # |
Yeah, I'm seeing the same problem on some images but not others, so my best guess is some kind of race. I'm not sure how best to proceed on that. We no longer do apt changes in the late command, unlike desktop, so I'm not sure where this change needs to happen.

LGTM. Thanks!