Hi Vasiliy,
On Thursday, June 02, 2016 02:29:51 PM Vasiliy Tolstov wrote:
> 2016-06-02 8:13 GMT+03:00 Jim Fehlig :
> > We've noticed libvirt-tck test 100-apply-verify-host.t failing recently on
> > libvirt.git master and I finally got around to bisecting it to commit
> >
Hi Daniel,
On Thursday, October 08, 2015 10:20:48 AM Daniel P. Berrange wrote:
> On Thu, Oct 08, 2015 at 10:17:23AM +0100, Daniel P. Berrange wrote:
> > On Mon, Oct 05, 2015 at 05:18:42PM -0600, Mike Latimer wrote:
> > > diag "Creating a new transient domain";
>
(sparse raw) and ~200K (qcow2)).
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
scripts/domain/121-block-info.t | 13 -
1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/scripts/domain/121-block-info.t b/scripts/domain/121-block-info.t
index e7e3491..64eeef2
All image types (allocated raw, sparse raw and qcow2) should be tested. Add
qcow2 to the test matrix, and reorder for clarify.
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
scripts/domain/121-block-info.t | 16 +++-
1 file changed, 11 insertions(+), 5 deletions(-)
diff
To avoid confusion between 'dst' and 'dst2' type of variables, rename
all variables relating to the fully allocated raw volume to include
the number '1'. (e.g. dst -> dst1, path -> path1, vol -> vol1, etc...)
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
scripts/domain/1
Hi,
Due to libvirt commit 0282ca45, 121-block-info.t fails as the allocation size
returned by block-info is now zero. While addressing this, a few aspects of
the test were also cleaned up. The changes are not extensive, but it made more
sense to submit them in a series.
Thanks,
Mike
Mike
changes
introduced in libvirt-tck commit 83ef279c.
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
lib/Sys/Virt/TCK/NetworkHelpers.pm| 2 +-
scripts/nwfilter/100-ping-still-working.t | 2 +-
scripts/nwfilter/210-no-mac-spoofing.t| 2 +-
scripts/nwfilter/220-no-ip-spoo
of the same
process owned by the dnsmasq user. This patch resolves the problem by adding
`uniq` to the command generating the process list.
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
scripts/networks/networkxml2hostout/tck-testnet-1.dat | 2 +-
scripts/networks/networkxml2hostout/tck-tes
Libvirt commit 0f87054b adds the --leasefile-ro option to dnsmasq. This change
results in the default.leases no longer containing the ip address for the test
domain. Instead of using the leases file, use arp to provide the ip address.
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
l
is restarted and the NUMA settings are compared to the
default flags (stored prior to any AFFECT_LIVE changes).
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
scripts/domain/202-numa-set-parameters.t | 14 +-
1 file changed, 9 insertions(+), 5 deletions(-)
diff --git a/s
When testing larger domains, a delay is necessary to ensure the test domain
is fully initialized before hotplugging disks and/or nics.
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
scripts/domain/200-disk-hotplug.t | 1 +
scripts/domain/210-nic-hotplug.t | 1 +
scripts/doma
Qemu removed support for the cow format in 550830f9. This patch drops all
cow related tests in libvirt-tck.
Signed-off-by: Mike Latimer <mlati...@suse.com>
---
scripts/storage/100-create-vol-dir.t | 19 +--
scripts/storage/200-clone-vol-dir.t | 4 ++--
2 files chan
shortly, dealing with issues in 121-block-info.t.
Mike Latimer (6):
Retrieve ip address from arp instead of leases
nwfilter startup and shutdown delay
NUMA flags are exclusive
Remove cow format tests
Delay before hotplugging
Handle dnsmasq child processes
lib/Sys/Virt/TCK/NetworkHelpers.pm
Correct the order of parameters passed to skip to ensure the following
tests are only executed under qemu and lxc environments:
hooks/051-daemon-hook.t
hooks/052-domain-hook.t
---
scripts/hooks/051-daemon-hook.t | 2 +-
scripts/hooks/052-domain-hook.t | 2 +-
2 files changed, 2
If nwfilter/050-apply-verify-host.t fails to start a test VM, a typo
causes the word 'Domain' to be executed as a command after undefining
the VM:
./nwfilter2vmtest.sh: line 425: Domain: command not found
Also, if either one of the test VM fails to be created, the test
filter should be deleted
Despite running all subtests successfully, hooks/051-daemon-hook.t ends
with a return code of 141. This is due to the connection to tck not being
cleaned up properly after libvirtd is restarted in the middle of the test.
Ignoring the SIGPIPE and specifically undefining $tck allows the test to
When TCK cleans up a test domain, existing snapshots are deleted. However,
not all drivers (e.g. libxl) support snapshots. When such a driver is
involved, the following error is reported and the testkit fails to cleanup
the domain:
libvirt error code: 3, message: this function is not supported
Under some environments (such as Xen), there may not be a uri_transport
defined. As this is not required, ensure the value is defined before
attempting to use it
---
lib/Sys/Virt/TCK/Capabilities.pm | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git
On Thursday, January 22, 2015 08:55:07 AM Cedric Bosdonnat wrote:
Seems like the apparmor profile for libvirtd is pretty wide open, so I'm
not sure if there will be much of a difference between those two
settings. I'm also not sure how best to test the functionality of those
helpers to
On Tuesday, January 20, 2015 09:08:04 AM Cedric Bosdonnat wrote:
On Mon, 2015-01-19 at 18:25 -0700, Mike Latimer wrote:
Apparmor must not prevent access to required helper programs. The
following
helpers should be allowed to run in unconfined execution mode:
- libvirt_parthelper
The network and nwfilter tests contained in the libvirt-TCK testkit can fail
unless access to raw network packets is granted. Without this access, the
following apparmor error can be seen while running the tests:
apparmor=DENIED operation=create parent=1 profile=/usr/sbin/libvirtd
pid=94731
Apparmor must not prevent access to required helper programs. The following
helpers should be allowed to run in unconfined execution mode:
- libvirt_parthelper
- libvirt_iohelper
---
examples/apparmor/usr.sbin.libvirtd | 2 ++
1 file changed, 2 insertions(+)
diff --git
In order for apparmor to work properly in Xen environments, the following
access rights need to be allowed:
- Allow CAP_SYS_PACCT, which is required when resetting some multi-port
Broadcom cards by writting to the PCI config space
- Allow CAP_IPC_LOCK, which is required to lock/unlock
.
Mike Latimer (3):
Fix apparmor issues for Xen
Grant access to helpers
Fix apparmor issues for tck
examples/apparmor/usr.sbin.libvirtd | 6 ++
1 file changed, 6 insertions(+)
--
1.8.4.5
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir
On Tuesday, April 22, 2014 11:35:31 AM Daniel P. Berrange wrote:
Unfortunately this is a change in API behaviour that affects more
than just the TCK. You need todo the equivalent change to the TCK
code instead really.
Ok. I didn't really view this as a change to the API, but I suppose it is.
Many libvirt-tck tests do not explicitly cleanup test domains before exiting.
In this case Sys::Virt::TCK-cleanup is triggered, and the environment is
reset automatically. During this reset, any existing snapshots are found for
deletion through Sys::Virt::Domain-list_snapshots. If the underlying
The 202-numa-set-parameters.t test sets NUMA_NODESET using AFFECT_LIVE on a
running domain, destroys and starts the domain, then verifies the NUMA_NODESET
setting. As AFFECT_LIVE does not write the setting to the domain xml file,
the new nodeset setting is lost when the domain is destroyed and the
With $tck added to the skip conditional, the entire codeblock has to be
after $tck is defined. Also, $tck-cleanup should be added to the skip
conditions.
---
scripts/nwfilter/300-vsitype.t | 18 ++
1 file changed, 10 insertions(+), 8 deletions(-)
diff --git
Trivial, but there are only 5 tests in 110-static-relabel-yes.t.
---
scripts/selinux/110-static-relabel-yes.t | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/scripts/selinux/110-static-relabel-yes.t
b/scripts/selinux/110-static-relabel-yes.t
index f558cc9..76781ca 100644
---
For environments not using a /24 netmask, this test can end up in a hung
state. This patch reads the netmask from the nic and uses it later when
ip addresses are changed and restored.
---
scripts/nwfilter/220-no-ip-spoofing.t | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff
Some environments (openSUSE 13.1) can report the network is unreachable during
this test. Trap that condition as well.
---
scripts/nwfilter/210-no-mac-spoofing.t | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/scripts/nwfilter/210-no-mac-spoofing.t
On Friday, March 28, 2014 11:09:02 AM Mike Latimer wrote:
If ebtables binary is not found in /sbin, use /usr/sbin.
---
scripts/nwfilter/100-ping-still-working.t | 3 ++-
scripts/nwfilter/210-no-mac-spoofing.t| 3 ++-
scripts/nwfilter/220-no-ip-spoofing.t | 3 ++-
scripts/nwfilter
On Friday, March 28, 2014 11:28:19 AM Mike Latimer wrote:
Using a statically defined vnet0 may fail if multiple VMs are running on the
test machine. Switch to mac address, but replace '00' with '0' to match the
output of ebtables.
---
scripts/nwfilter/100-ping-still-working.t | 4 ++--
1
On Friday, March 28, 2014 12:26:31 PM Daniel P. Berrange wrote:
Change tests which need a full OS image over to use the new
virtbuilder images instead of provisioning from a kickstart
file. This should make them much more reliable to run.
Signed-off-by: Daniel P. Berrange berra...@redhat.com
On Friday, March 28, 2014 12:26:26 PM Daniel P. Berrange wrote:
A bunch of tests currently attempt to kickstart a full Fedora
OS image install. Everytime I try to update this kickstart to
a new version of Fedora it causes no end of pain. Switch the
tests over to use Richard Jones' virt-builder
On Friday, March 28, 2014 03:16:37 PM Guido Günther wrote:
On Fri, Mar 28, 2014 at 12:26:26PM +, Daniel P. Berrange wrote:
A bunch of tests currently attempt to kickstart a full Fedora
OS image install. Everytime I try to update this kickstart to
a new version of Fedora it causes no end
If ebtables binary is not found in /sbin, use /usr/sbin.
---
scripts/nwfilter/100-ping-still-working.t | 3 ++-
scripts/nwfilter/210-no-mac-spoofing.t| 3 ++-
scripts/nwfilter/220-no-ip-spoofing.t | 3 ++-
scripts/nwfilter/230-no-mac-broadcast.t | 3 ++-
On Friday, March 28, 2014 05:31:13 PM Richard W.M. Jones wrote:
libguestfs exists on SUSE. Olaf Hering is packaging it. Not sure
about SLES, but I was under the impression it was packaged for at
least some version of SLES.
Right. I thought he was working on some virt-builder images as well,
Using a statically defined vnet0 may fail if multiple VMs are running on the
test machine. Switch to mac address, but replace '00' with '0' to match the
output of ebtables.
---
scripts/nwfilter/100-ping-still-working.t | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git
On Friday, March 28, 2014 05:34:19 PM Richard W.M. Jones wrote:
On Fri, Mar 28, 2014 at 11:06:56AM -0600, Mike Latimer wrote:
ACK to all of this. For some reason, Fedora 20 didn't want to install for
me (it failed when resizing the disk, and I don't have time to chase it
today), but Fedora
On Friday, March 28, 2014 02:32:28 PM Daniel P. Berrange wrote:
I already made it configurable - just change the OS name in the config
file to any that virt-builder supports.
Even better. I'm very much liking what I'm seeing in the patches -
specifically patch 4/6. I'll try to test it today
On Friday, March 28, 2014 12:26:26 PM Daniel P. Berrange wrote:
A bunch of tests currently attempt to kickstart a full Fedora
OS image install. Everytime I try to update this kickstart to
a new version of Fedora it causes no end of pain. Switch the
tests over to use Richard Jones' virt-builder
On Friday, March 28, 2014 12:28:43 PM Daniel P. Berrange wrote:
On Thu, Mar 27, 2014 at 04:12:16PM -0600, Mike Latimer wrote:
Move the test for /usr/sbin/lldptool up so libvirt-tck will report the
skip
and reason, rather than passing the test as 'ok'.
So we're talking about
On Thursday, March 27, 2014 11:45:06 AM Daniel P. Berrange wrote:
When printing test results, nwfilter2vmtest.sh, truncates
the output at 66 or 70 characters. This is very unhelpful
when debugging problems where you want to see the full
output
Signed-off-by: Daniel P. Berrange
On Thursday, March 27, 2014 11:10:15 AM Eric Blake wrote:
New or not, your reviews are helpful! We all had to start somewhere.
And right now, the fact that you are attempting to run tck gives you
better status as a reviewer, even if only to say I applied this, and it
{still passed for
On Thursday, March 27, 2014 02:49:11 PM Daniel P. Berrange wrote:
To make it easier to add more optional parameters to the
Sys::Virt::TCK::generic_domain method, change it to take
named parameters instead of positional parameters.
Signed-off-by: Daniel P. Berrange berra...@redhat.com
---
On Thursday, March 27, 2014 03:38:47 PM Daniel P. Berrange wrote:
The 202-numa-set-parameters.t test would attempt to fetch
NUMA parameters while the guest is shutoff using the
VIR_DOMAIN_AFFECT_LIVE parameter. Change it to use the
param AFFECT_CONFIG instead, and add a separate test
for
On Thursday, March 27, 2014 02:49:25 PM Daniel P. Berrange wrote:
The network tests invoke various ifconfig and route commands
to test network setup, and also grep for dnsmasq/radvd args.
Switch to use 'ip' since ifconfig and route commands are not
installed by default on recent distros any
Move the test for /usr/sbin/lldptool up so libvirt-tck will report the skip
and reason, rather than passing the test as 'ok'.
---
scripts/nwfilter/300-vsitype.t | 48 ++
1 file changed, 25 insertions(+), 23 deletions(-)
diff --git
Due to iptables commit 945353a2 (in iptables v1.4.20 and higher), ip6tables
no longer prints out /128. This patch removes /128 from output files, and
replaces '/128' in command output with '' to remain compatible with
older versions of ip6tables.
---
Hi,
As I've been looking through libvirt-tck tests, I found that commit 4f209434
(in libvirt) changes a condition that the nwfilter/050-apply-verify-host.t
relies on.
Specifically, the 050-apply-verify-host.t test creates a number of filters with
invalid values (such as dscp='64', and
06, 2014 03:41:01 PM Mike Latimer wrote:
Reload libvirtd after hook testing has completed. Otherwise, libvirtd
is still expecting hook test scripts to exist.
---
scripts/hooks/052-domain-hook.t | 4
1 file changed, 4 insertions(+)
diff --git a/scripts/hooks/052-domain-hook.t
b
The network/100-apply-verify-host.t test compares the results of various
network commands (route, iptables, ebtables, etc...) against known, valid
results. Depending on the versions of these tools, minor whitespace
differences in command output can result in the failure of the test. This
patch
Ping. Has anyone (possibly Osier) had a chance to review this patch?
On Friday, March 07, 2014 11:06:35 AM Mike Latimer wrote:
When matching capabilities of a guest, if multiple domain types exist (for
example, 'qemu' and 'kvm') the order in which they are returned can change.
To avoid
The main libvirt-tck script only executes tests using the default (.t)
extension. This patch adds .t and .pl as valid extensions using the --ext
parameter. This is required to enable storage tests added in commit 97c6a506.
---
Build.PL| 2 +-
bin/libvirt-tck | 5 +
2 files changed, 6
Commit 97c6a506 added several storage tests with a .pl extension. Rather than
enable .pl tests, this patch simply renames these tests to .t.
---
scripts/storage/{400-vol-download.pl = 400-vol-download.t} | 0
scripts/storage/{405-vol-download-all.pl = 405-vol-download-all.t}
On Monday, March 24, 2014 04:55:38 PM Daniel P. Berrange wrote:
On Mon, Mar 24, 2014 at 10:42:35AM -0600, Mike Latimer wrote:
The main libvirt-tck script only executes tests using the default (.t)
extension. This patch adds .t and .pl as valid extensions using the --ext
parameter
On Monday, March 24, 2014 02:24:23 PM Eric Blake wrote:
'sed -i' is not portable (it is a GNU-ism) - but it's not the first time
we've assumed GNU tools in this testsuite so I can overlook it.
Ok. I'll keep that in mind in the future. If you'd rather the -i not be used,
it's easy enough to
Sorry for the slow response, I've been distracted with other projects.
On Monday, March 10, 2014 09:25:37 PM Osier Yang wrote:
Hm, $tck-cleanup() doesn't close the connection, it just destroy and
undefine the existing domains, networks, and pools.
snip
sub reset {
my $self = shift;
On Friday, March 07, 2014 08:54:26 AM Eric Blake wrote:
On 03/07/2014 02:02 AM, Osier Yang wrote:
On 07/03/14 00:27, Mike Latimer wrote:
Starting with libvirt 0.9.0+, hook scripts can be called from several new
locations. These locations must also be reflected in the expected logs
On Friday, March 07, 2014 05:02:00 PM Osier Yang wrote:
ACK, but wondering why we didn't discover it, according to the log for
the hook scripts have been changed.
I'm not sure why this hasn't been seen before. The only thing that makes sense
is that the hook tests haven't been tested since the
On Friday, March 07, 2014 05:16:48 PM Osier Yang wrote:
$hook-cleanup();
+
+# Restarting libvirtd broke the tck connection, so ignore sigpipe and
+# undefine $tck to avoid a return code of 141
+$SIG{PIPE} = 'IGNORE';
+undef $tck;
We should get the libvirt
On Friday, March 07, 2014 05:55:34 PM Osier Yang wrote:
- return ($domains[0],
+# Prefer kvm if multiple domain types are returned
+my $domain;
+if (int(@domains) gt 1) {
+for (my $j = 0 ; $j int(@domains) ; $j++) {
I would use grep instead
When matching capabilities of a guest, if multiple domain types exist (for
example, 'qemu' and 'kvm') the order in which they are returned can change.
To avoid unpredictable test results, this patch prefers kvm if that domain
type exists. If not, the behavior matches what existed before, and the
Starting with libvirt 0.9.0+, hook scripts can be called from several new
locations. These locations must also be reflected in the expected logs of
the hook tests.
The final test in 052-domain-hook.t intentionally produces a failed start,
which should show the first stage in the startup hook
This test completes successfully, but results in a return code of 141 due to
a broken pipe when restarting libvirtd. This patch just masks the SIGPIPE
and undefines $tck to avoid the 141 return code. If there is a way to
reestablish the tck connection after the restart, that would be a better
When matching capabilities of a guest, if multiple domain types exist (for
example, 'qemu' and 'kvm') the order in which they are returned can change.
To avoid unpredictable test results, this patch prefers kvm if that domain
type exists. If not, the behavior matches what existed before, and the
Reload libvirtd after hook testing has completed. Otherwise, libvirtd
is still expecting hook test scripts to exist.
---
scripts/hooks/052-domain-hook.t | 4
1 file changed, 4 insertions(+)
diff --git a/scripts/hooks/052-domain-hook.t b/scripts/hooks/052-domain-hook.t
index
68 matches
Mail list logo