CRIU's pre-copy migration support relies on the soft dirty page tracking
in the Linux kernel:
https://www.kernel.org/doc/Documentation/vm/soft-dirty.txt
This functionality is not implemented for all architectures and it can
also be turned off in the kernel.
CRIU can check if the combination of architecture/kernel/CRIU supports
the soft dirty page tracking and exports this feature checking
functionality in go-criu.
This commit adds an early check if the user selects pre-copy
checkpointing to error out if the system does not support it.
Signed-off-by: Adrian Reber <areber@redhat.com>
This extends one of the checkpoint/restore tests to see if the
newly introduced checkpoint specific fields in 'inspect' work as
intended.
Signed-off-by: Adrian Reber <areber@redhat.com>
Write a BeValidJSON() matcher, and replace IsJSONOutputValid():
sed -i -e 's/Expect(\(.*\)\.IsJSONOutputValid()).To(BeTrue())/Expect(\1.OutputToString())\.To(BeValidJSON())/' test/e2e/*_test.go
(Plus a few manual tweaks)
Signed-off-by: Ed Santiago <santiago@redhat.com>
Many ginkgo tests have been written to use this evil form:
GrepString("foo")
Expect(that to BeTrue())
...which yields horrible useless messages on failure:
false is not true
Identify those (automatically, via script) and convert to:
Expect(output to ContainSubstring("foo"))
...which yields:
"this output" does not contain substring "foo"
There are still many BeTrue()s left. This is just a start.
This is commit 1 of 2. It includes the script I used, and
all changes to *.go are those computed by the script.
Commit 2 will apply some manual fixes.
Signed-off-by: Ed Santiago <santiago@redhat.com>
On container restore ensures that the same container runtime is used as
during checkpointing and it also ensures that the user does not select
a different runtime.
Signed-off-by: Adrian Reber <areber@redhat.com>
* Support `checkpoint --pre-checkpoint`
* Support `checkpoint --with-previous`
* Disable `restore --import-previous` for the remote client since we had
to send two files which in turn would require to tar them up and hence
be a breaking change. Podman 4.0 would be the chance and I hope we'll
find time before that to remote-restore prettier.
Note that I did not run over swagger yet to check whether all parameters
are actually documented due to time constraints.
Fixes: #12334
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
Nothing was working before, and it's too much to summarize. To make
sure we're not regressing in the future again, enable the remote e2e
tests.
Fixes: #12007
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
Trying to restore a container that was started with '--ipc host' fails
with:
Error: error creating container storage: ProcessLabel and Mountlabel must either not be specified or both specified
We already fixed this exact same error message for containers started
with '--privileged'. The previous fix was to check if the to be restored
container is a privileged container (c.config.Privileged). Unfortunately
this does not work for containers started with '--ipc host'.
This commit changes the check for a privileged container to check if
both the ProcessLabel and the MountLabel is actually set and only then
re-uses those labels.
Signed-off-by: Adrian Reber <areber@redhat.com>
It was observed during initial F35 testing, this test can cause Ginkgo
to "hang" by attempting to connect before the redis is up/listening.
Fix this by confirming the ready-state before attempting to connect.
Also, force IPv4 and timeout on any connection fault - to allow other
tests to run.
Thanks to Adrian Reber for help on this and related fixes.
Signed-off-by: Chris Evich <cevich@redhat.com>
Moving to Fedora 35 showed test failures (time outs) in the test
"podman checkpoint and restore container with different port mappings"
The test starts a container and maps the internal port 6379 to the local
port 1234 ('-p 1234:6379') and then tries to connect to localhost:1234
On Fedora 35 this failed and blocked the test because the container was
not yet ready. The test was trying to connect to localhost:1234 but
nothing was running there. So the error was not checkpointing related.
Before trying to connect to the container the test is now waiting for
the container to be ready.
Another problem with this test and running ginkgo in parallel was that
it was possible that the port was already in use. Now for each run a
random port is selected to decrease the chance of collisions.
Signed-off-by: Adrian Reber <areber@redhat.com>
Add --time flag to podman container rm
Add --time flag to podman pod rm
Add --time flag to podman volume rm
Add --time flag to podman network rm
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
When inspecting a container, we now report whether the container
was stopped by a `podman checkpoint` operation via a new bool in
the State portion of inspected, `Checkpointed`.
Signed-off-by: Matthew Heon <mheon@redhat.com>
The upcoming commit to support checkpointing out of Pods requires CRIU
3.16. This changes the CRIU version check to support checking for
different versions.
Signed-off-by: Adrian Reber <areber@redhat.com>
e2e test failures are rife with messages like:
Expected 1 to equal 0
These make me cry. They're anti-helpful, requiring the reader
to dive into the source code to figure out what those numbers
mean.
Solution: Go tests have a '.Should(Exit(NNN))' mechanism. I
don't know if it spits out a better diagnostic (I have no way
to run e2e tests on my laptop), but I have to fantasize that
it will, and given the state of our flakes I assume that at
least one test will fail and give me the opportunity to see
what the error message looks like.
THIS IS NOT REVIEWABLE CODE. There is no way for a human
to review it. Don't bother. Maybe look at a few random
ones for sanity. If you want to really review, here is
a reproducer of what I did:
cd test/e2e
! positive assertions. The second is the same as the first,
! with the addition of (unnecessary) parentheses because
! some invocations were written that way. The third is BeZero().
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.To\(Equal\((\d+)\)\)/Expect($1).Should(Exit($2))/' *_test.go
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.To\(\(Equal\((\d+)\)\)\)/Expect($1).Should(Exit($2))/' *_test.go
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.To\(BeZero\(\)\)/Expect($1).Should(Exit(0))/' *_test.go
! Same as above, but handles three non-numeric exit codes
! in run_exit_test.go
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.To\(Equal\((\S+)\)\)/Expect($1).Should(Exit($2))/' *_test.go
! negative assertions. Difference is the spelling of 'To(Not)',
! 'ToNot', and 'NotTo'. I assume those are all the same.
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.To\(Not\(Equal\((0)\)\)\)/Expect($1).To(ExitWithError())/' *_test.go
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.ToNot\(Equal\((0)\)\)/Expect($1).To(ExitWithError())/' *_test.go
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.NotTo\(Equal\((0)\)\)/Expect($1).To(ExitWithError())/' *_test.go
! negative, old use of BeZero()
perl -pi -e 's/Expect\((\S+)\.ExitCode\(\)\)\.ToNot\(BeZero\(\)\)/Expect($1).Should(ExitWithError())/' *_test.go
Run those on a clean copy of main branch (at the same branch
point as my PR, of course), then diff against a checked-out
copy of my PR. There should be no differences. Then all you
have to review is that my replacements above are sane.
UPDATE: nope, that's not enough, you also need to add gomega/gexec
to the files that don't have it:
perl -pi -e '$_ .= "$1/gexec\"\n" if m!^(.*/onsi/gomega)"!' $(grep -L gomega/gexec $(git log -1 --stat | awk '$1 ~ /test\/e2e\// { print $1}'))
UPDATE 2: hand-edit run_volume_test.go
UPDATE 3: sigh, add WaitWithDefaultTimeout() to a couple of places
UPDATE 4: skip a test due to bug #10935 (race condition)
Signed-off-by: Ed Santiago <santiago@redhat.com>
We missed bumping the go module, so let's do it now :)
* Automated go code with github.com/sirkon/go-imports-rename
* Manually via `vgrep podman/v2` the rest
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
This implements support for mounting and unmounting volumes
backed by volume plugins. Support for actually retrieving
plugins requires a pull request to land in containers.conf and
then that to be vendored, and as such is not yet ready. Given
this, this code is only compile tested. However, the code for
everything past retrieving the plugin has been written - there is
support for creating, removing, mounting, and unmounting volumes,
which should allow full functionality once the c/common PR is
merged.
A major change is the signature of the MountPoint function for
volumes, which now, by necessity, returns an error. Named volumes
managed by a plugin do not have a mountpoint we control; instead,
it is managed entirely by the plugin. As such, we need to cache
the path in the DB, and calls to retrieve it now need to access
the DB (and may fail as such).
Notably absent is support for SELinux relabelling and chowning
these volumes. Given that we don't manage the mountpoint for
these volumes, I am extremely reluctant to try and modify it - we
could easily break the plugin trying to chown or relabel it.
Also, we had no less than *5* separate implementations of
inspecting a volume floating around in pkg/infra/abi and
pkg/api/handlers/libpod. And none of them used volume.Inspect(),
the only correct way of inspecting volumes. Remove them all and
consolidate to using the correct way. Compat API is likely still
doing things the wrong way, but that is an issue for another day.
Fixes#4304
Signed-off-by: Matthew Heon <matthew.heon@pm.me>
Add some more tests, document cases where remote will not work
Add FIXMEs for tests that should work on podman-remote but currently
do not.
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
With the advent of Podman 2.0.0 we crossed the magical barrier of go
modules. While we were able to continue importing all packages inside
of the project, the project could not be vendored anymore from the
outside.
Move the go module to new major version and change all imports to
`github.com/containers/libpod/v2`. The renaming of the imports
was done via `gomove` [1].
[1] https://github.com/KSubedi/gomove
Signed-off-by: Valentin Rothberg <rothberg@redhat.com>
Podman was checking if the runtime support checkpointing by running
'runtime checkpoint -h'. That works for runc.
crun, however, does not use '-h, --help' for help output but, '-?,
--help'.
This commit switches both checkpoint support detection from
'runtime checkpoint -h'
to
'runtime checkpoint --help'.
Podman can now correctly detect if 'crun' also support checkpointing.
Signed-off-by: Adrian Reber <areber@redhat.com>
When doing a checkpoint with --export the root file-system diff was not
working as expected. Instead of getting the changes from the running
container to the highest storage layer it got the changes from the
highest layer to that parent's layer. For a one layer container this
could mean that the complete root file-system is part of the checkpoint.
With this commit this changes to use the same functionality as 'podman
diff'. This actually enables to correctly diff the root file-system
including tracking deleted files.
This also removes the non-working helper functions from libpod/diff.go.
Signed-off-by: Adrian Reber <areber@redhat.com>
Trying to checkpoint a container started with --rm works, but it makes
no sense as the container, including the checkpoint, will be deleted
after writing the checkpoint. This commit inhibits checkpointing
containers started with '--rm' unless '--export' is used. If the
checkpoint is exported it can easily be restored from the exported
checkpoint, even if '--rm' is used. To restore a container from a
checkpoint it is even necessary to manually run 'podman rm' if the
container is not started with '--rm'.
Signed-off-by: Adrian Reber <areber@redhat.com>
Rather than checking for non-zero, we need to check for >0 to
distinguish between timeouts and error exit codes.
Signed-off-by: Jhon Honce <jhonce@redhat.com>
In the restore from external checkpoint archive test, the second restore
using a new name and ID is now done first to ensure that nothing in the
restored container depends on the original container.
Test has been adapted to catch errors like the one fixed with the
previous commit to adapt ConmonPidFile for restored containers.
Signed-off-by: Adrian Reber <areber@redhat.com>
If a container is restored multiple times from an exported checkpoint
with the help of '--import --name', the restore will fail if during
'podman run' a static container IP was set with '--ip'. The user can
tell the restore process to ignore the static IP with
'--ignore-static-ip'.
Signed-off-by: Adrian Reber <areber@redhat.com>
This enables programs and scripts wrapping the podman command to handle
'podman rm' and 'podman rmi' failures caused by paused or running
containers or due to images having other child images or dependent
containers. These errors are common enough that it makes sense to have
a more machine readable way of detecting them than parsing the standard
error output.
Signed-off-by: Ondrej Zoder <ozoder@redhat.com>
Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
The function to generate random IP addresses during ginkgo tests in
the checkpoint test code is moved to common and all tests using
hardcoded IP addresses have been changed to use random IP addresses to
reduce test errors when running the tests in parallel.
Signed-off-by: Adrian Reber <areber@redhat.com>
This adds three tests for the --ignore-rootfs option to verify that it
works in all combination.
1. Not used at all
2. Only used during restore
3. Only used during checkpoint
Signed-off-by: Adrian Reber <areber@redhat.com>
Valentin Rothberg
Adrian Reber
Radostin Stoyanov
Ed Santiago
Radostin Stoyanov
OpenShift Merge Robot
Chris Evich
Daniel J Walsh
Matthew Heon
Matthew Heon
unknown
Brent Baude
Jakub Filak
Jhon Honce