Dimitri John Ledkov
5a72d361a7
Merge pull request #576 from mwhudson/unmount-before-probing
...
if /target exists, run curtin umount /target before first disk probe
2019-11-18 03:40:25 -08:00
Michael Hudson-Doyle
ec14875a85
if /target exists, run curtin umount /target before first disk probe
...
The installer now lets you restart after an install failure, but unless
we do something like this, the target system still being mounted will
make life confusing.
It's more robust to be defensive and try to unmount at startup rather
than trying to unmount before restarting.
2019-11-18 15:35:56 +13:00
Michael Hudson-Doyle
ac071ec9bf
Merge pull request #574 from mwhudson/error-report-ui
...
first cut at error report ui
2019-11-18 12:30:50 +13:00
Michael Hudson-Doyle
5c509511f3
read mountinfo in one go
2019-11-18 12:30:17 +13:00
Michael Hudson-Doyle
330bbeb024
Merge pull request #575 from mwhudson/no-cache-storage-probe-data
...
stop caching probed storage data
2019-11-13 10:32:02 +13:00
Michael Hudson-Doyle
73b53f2bf3
stop caching probed storage data
...
A recent branch carefully called Prober.get_storage() whenever a udev
block device event was seen. But Prober.get_storage() cached the results
from probert rendering that useless. Rip the caching and some other
complexity out of the prober class.
2019-11-13 10:07:51 +13:00
Michael Hudson-Doyle
b5ea14f4f1
fake persistent_details in dry-run mode
2019-11-12 11:42:28 +13:00
Michael Hudson-Doyle
d550bb139d
design inspired tweaks
2019-11-12 11:33:38 +13:00
Michael Hudson-Doyle
d926cc5178
show the report for an install failure when it happens
2019-11-12 11:32:50 +13:00
Michael Hudson-Doyle
0affdaf046
present kind-appropriate options when an error occurs
2019-11-12 11:32:22 +13:00
Michael Hudson-Doyle
3207170da5
show probe failures on reaching the filesystem screen
2019-11-12 11:31:11 +13:00
Michael Hudson-Doyle
615dc1b1dd
show an unseen ui crash on startup
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
009d0598d0
update error list when report changes (esp. is viewed)
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
3221b14294
load crash reports from disk on startup
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
3837780b60
show persistent location of crash report
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
1d413b6b06
error list view
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
e93b5111b3
update error view when report is done
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
50e9dfb324
show error reports on creation (mostly)
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
6d3f014f7a
simple view of a crash report
2019-11-08 13:36:30 +13:00
Michael Hudson-Doyle
a8c1be5b2c
Merge pull request #573 from mwhudson/raid-size-calculation
...
More accurate estimation of the size of a RAID
2019-11-08 10:12:43 +13:00
Michael Hudson-Doyle
0141defa01
changes suggested in review
2019-11-08 09:47:34 +13:00
Michael Hudson-Doyle
65a961be1a
do not run the 100T raid tests, mdadm fails with SIGPIPE
2019-11-08 09:31:51 +13:00
Michael Hudson-Doyle
5d8c983448
make raid-size-tests.py a tiny bit friendlier
2019-11-08 09:31:51 +13:00
Michael Hudson-Doyle
d532648794
beef up comments
2019-11-08 09:31:49 +13:00
Michael Hudson-Doyle
fde1c53141
sort raid devices consistently when computing size and when serializing
2019-11-08 09:30:22 +13:00
Michael Hudson-Doyle
233965b376
fix raid size calculation when sizes of devices in array vary
...
It turns out mdadm computes the headroom based on the first device in
the array, which means that the order of the devices matters!
2019-11-08 09:22:59 +13:00
Michael Hudson-Doyle
6b5e7adcf6
add some explanation to raid-size-tests.py
2019-11-08 09:22:59 +13:00
Michael Hudson-Doyle
5da663fede
raid-size-tests.py: run in tmpfs, clean up better
...
ext4 limits files to a measly 16TiB!
2019-11-08 09:22:59 +13:00
Michael Hudson-Doyle
679b611077
add tests of get_raid_size against reality
2019-11-08 09:22:59 +13:00
Michael Hudson-Doyle
53e532e602
add a test from the only bug i can find where the size of the resulting raid is reported
2019-11-08 09:22:59 +13:00
Michael Hudson-Doyle
ea9af1cbc0
More accurate estimation of the size of a RAID
...
I found the code in mdadm that decides how much of a device is usable
for data and copied it. It's a bit hairy!
For https://bugs.launchpad.net/subiquity/+bug/1816777 and many similar
bugs.
2019-11-08 09:22:59 +13:00
Michael Hudson-Doyle
2baf9583db
Merge pull request #569 from mwhudson/error-report-5
...
generate crash reports
2019-11-07 15:39:19 +13:00
Michael Hudson-Doyle
7908a958d0
create crash reports
2019-11-07 14:47:20 +13:00
Michael Hudson-Doyle
74ef2ad8f6
add a way to create a crash report
2019-11-07 14:47:18 +13:00
Michael Hudson-Doyle
38dfeebde7
Add ErrorReport and ErrorController classes
2019-11-07 14:46:58 +13:00
Michael Hudson-Doyle
7691e26489
add and use a way for code to note data that should end up in a crash report
2019-11-07 14:46:57 +13:00
Dimitri John Ledkov
dd1bcc9a89
Merge pull request #570 from mwhudson/monitor-udev
...
reprobe block devices on udev events
2019-11-06 14:04:06 +00:00
Michael Hudson-Doyle
2ec5b3b2d4
Merge pull request #572 from mwhudson/bump-probert
...
bump probert
2019-11-06 22:27:13 +13:00
Michael Hudson-Doyle
4afb53a7ce
bump probert
2019-11-06 14:59:25 +13:00
Michael Hudson-Doyle
4d5c217063
Merge pull request #571 from mwhudson/curtin-ubuntu-devel
...
go back to getting curtin from ubuntu/devel branch
2019-11-06 10:43:18 +13:00
Michael Hudson-Doyle
b0d19fc58b
go back to getting curtin from ubuntu/devel branch
2019-11-06 10:34:37 +13:00
Michael Hudson-Doyle
69c5e8179f
fix logic around waiting for udev to settle
2019-11-05 10:39:06 +13:00
Michael Hudson-Doyle
54823e364c
remove signal sent after debug shell exits
2019-11-04 23:52:49 +13:00
Michael Hudson-Doyle
4fa941ef4a
reprobe block devices on udev block events
2019-11-04 23:50:00 +13:00
Michael Hudson-Doyle
4798348740
Merge pull request #567 from mwhudson/restricted-saved-probing
...
implement restricted probing for canned probert data
2019-11-04 10:44:18 +13:00
Michael Hudson-Doyle
bd7f0eef1d
Merge pull request #568 from mwhudson/controlled-failures
...
add a controlled way to make various things fail
2019-11-04 10:43:48 +13:00
Michael Hudson-Doyle
ca02972195
add a controlled way to make various things fail
...
In dry-run mode, treat SUBIQUITY_DEBUG as a comma-separated list of
things to fail:
* install-fail: curtin install fails
* copy-logs-fail: the postinstall copying of logs fails
* bpfail-full: full block device probing fails
* bpfail-restricted: restricted block device probing fails
In addition, control-u makes the UI crash in dry-run mode.
(This is all for testing of crash report generation but does not depend
on that).
2019-11-04 10:43:29 +13:00
Michael Hudson-Doyle
d56b72b35b
implement restricted probing for canned probert data
2019-10-31 13:39:01 +13:00
Michael Hudson-Doyle
3d1fb15dae
Merge pull request #566 from CanonicalLtd/drop-deb-packages
...
Drop crufty .deb packages from the source package build
2019-10-31 09:42:19 +13:00
Steve Langasek
f6416abd78
Don't call dh_missing
...
Just don't bother, it's not worth creating a whitelist that's 3x longer than
the actual target packages we're building
2019-10-30 10:57:54 -07:00