Extend ssh-server QA with new test cases for pubkey authentication thus
covering `SSH.AuthorizedKey.` feature using ed25519 key algorithm,
furtherore add check for `SSH.Server.{i}.Session.{i}.` active SSH session
list feature.
Closes: PCF-1026
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently Device.Security.Certificate.{i}.Enable is read-only so we're
unable to test it until its fixed.
References: PCF-1054
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently we've different value for certificate's NotBefore/NotAfter
timestamps on each service startup which would result in not
reproducible test results, so lets ignore the microseconds part until
its fixed.
References: PCF-1053
Signed-off-by: Petr Štetiar <ynezz@true.cz>
It wasn't possible to change user passwords PCF-539, but this has been
fixed and thus we should cover this use case with a test to prevent
regressions.
Closes: PCF-542
Signed-off-by: Petr Štetiar <ynezz@true.cz>
We're using eMMC to boot the OSP boards, so lets make it clear and fix
the wrongly used copy&paste-d initramfs.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Adding obuspa package has increased slightly prplOS firmware image, but
it was enough to trigger Haze's bootloader factory pre-configured
limits.
U-Boot shipped with the devices has configuration limits which prevents
to boot initramfs images which have size of 33951166 Bytes = 32.4 MiB:
U-Boot 2016.01 (Nov 09 2022 - 16:51:22 +0800)
...
IPQ807x# tftpboot 0x41000000 openwrt-ipq807x-generic-prpl_haze-initramfs-uImage.itb && bootm 0x41000000
...
Uncompressing Kernel Image ... Error: inflate() returned -5
Image too large: increase CONFIG_SYS_BOOTM_LEN
Must RESET board to recover
resetting ...
So lets workaround this bootloader limit by writting the firmware image
directly into the eMMC storage with sysupgrade image, thats it, we
couldn't use initramfs images on Haze until we get U-Boot with suitable
configuration.
References: PCF-1048, PCF-929
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently there is no WAN connection over XPON interface, so until its
fixed, lets make the CDRouter tests manual.
References: PCF-982
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently its not possible to start any CDRouter test on Turris Omnia
device due to the following error:
FAIL: Could not establish IPv4 DHCP client on WAN interface » no-wan-dhcp-address
It was found out, that this is happening due to the crash of DHCPv4
client:
Core was generated by `tr181-dhcpv4client -D'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0 0xb6f44510 in memchr (src=src@entry=0xf, c=c@entry=0, n=<optimized out>, n@entry=2147483647) at src/string/memchr.c:16
16 src/string/memchr.c: No such file or directory.
(gdb) bt
#0 0xb6f44510 in memchr (src=src@entry=0xf, c=c@entry=0, n=<optimized out>, n@entry=2147483647) at src/string/memchr.c:16
#1 0xb6f45b90 in strnlen (s=s@entry=0xf <error: Cannot access memory at address 0xf>, n=2147483647) at src/string/strnlen.c:5
#2 0xb6f0b048 in printf_core (f=f@entry=0xbecc23f8, fmt=fmt@entry=0xb6bd953d " with this lease%s - %s(%s@%s:%d)%s", ap=ap@entry=0xbecc22f4, nl_arg=nl_arg@entry=0xbecc2320,
nl_type=<optimized out>, nl_type@entry=0xbecc22f8) at src/stdio/vfprintf.c:594
#3 0xb6f40014 in vfprintf (f=f@entry=0xbecc23f8, fmt=fmt@entry=0xb6bd953d " with this lease%s - %s(%s@%s:%d)%s", ap=..., ap@entry=...) at src/stdio/vfprintf.c:683
#4 0xb6f42784 in vsnprintf (s=s@entry=0xbecc253c "client: ", n=n@entry=984, fmt=fmt@entry=0xb6bd953d " with this lease%s - %s(%s@%s:%d)%s", ap=ap@entry=...) at src/stdio/vsnprintf.c:54
#5 0xb6f247c8 in _vsyslog (priority=<optimized out>, priority@entry=4, message=message@entry=0xb6bd953d " with this lease%s - %s(%s@%s:%d)%s", ap=ap@entry=...) at src/misc/syslog.c:108
#6 0xb6f24c74 in __vsyslog (priority=4, message=0xb6bd953d " with this lease%s - %s(%s@%s:%d)%s", message@entry=0xb6bd4944 "d)%s", ap=...) at src/misc/syslog.c:133
#7 0xb6d1a128 in sahTraceOutput (lvl=<optimized out>, traceLvl=<optimized out>, format=format@entry=0xb6bd4944 "d)%s", vl=...) at sahtrace.c:154
#8 0xb6d1a760 in sahTracev (lvl=<optimized out>, format=format@entry=0xb6bd4944 "d)%s", ap=..., ap@entry=...) at sahtrace.c:754
#9 0xb6d1a798 in sahTrace (lvl=<optimized out>, format=0xb6bd953d " with this lease%s - %s(%s@%s:%d)%s") at sahtrace.c:738
#10 0xb6bd0680 in dhcp_lease_config_req_ip_option (lease=<optimized out>) at dhcp_lease.c:368
#11 0xb6bd4944 in all_implied_fbits () from /home/builder/scripts/../staging_dir/target-arm_cortex-a9+vfpv3-d16_musl_eabi/root-mvebu/usr/lib/amx/tr181-dhcpv4client/modules/mod-dhcpv4c.so
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
So until its fixed, lets make the CDRouter tests manual.
References: PCF-1051
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Since we now have eMMC boot available on the Haze target, lets enable
TR-069 CDRouter testing on this target as well.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Add prpl-haze-emmc so we can flash prplOS into eMMC flash storage using
sysupgrade image directly from the U-Boot bootloader.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
New obuspa package was just added, thus we need to QA it on CI to
prevent regressions, so lets enable CDRouter testing stock using
CDRouter-USP-Certification package.
References: PCF-1005
Signed-off-by: Petr Štetiar <ynezz@true.cz>
New obuspa package was just added, thus we need to QA it on CI to prevent
regressions, so lets make CDRouter aware, that the CPEs support USP over
STOMP protocol.
References: PCF-1005
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Added OB-USP-A package to feed-prpl with default configuration tailored
for CDRouter's USP certification testing default settings.
To make USP available in the system we need to backport as well fix for
libamxrt which makes libamxrt to only connect to default backend sockets
when the necessary backend is loaded. So if the USP backend is not
loaded, the plugin won't connect to the USP sockets in /var/run/usp/.
Fixes: PCF-1067
References: PCF-1005
Signed-off-by: Sandesh Mahale <sandesh.mahale@commscope.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz> [libamxrt fix]
Currently the feed-prpl CI build testing pipeline suddenly fails:
$ pip3 install jinja2
error: externally-managed-environment
This environment is externally managed
╰─>
The system-wide python installation should be maintained using the system
package manager (apk) only.
...
Thats because we’re using latest Alpine Linux containers and those get
update to latest Alpine Linux 3.19.0 release just a few days ago on
December 7th, 2023 with following note related to the above in the
release notes:
Python’s package directory is now marked as externally managed, which
means that pip can no longer install to system directory which is
managed by apk.
So lets fix it by using jinja2 apk package directly as suggested.
Fixes: PCF-1055
References: https://www.alpinelinux.org/posts/Alpine-3.19.0-released.html
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Disabling prplmesh restart and remove prplmesh check test
because of PPM-2674
opertaing wireless test is impacted disable it also.
References: PPM-2674
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz> [facelift]
Disabling the restart of prplmesh and remove prplmesh check because
we've noticed on latest automated prplos test that when we run pwhm as
gateway mode (agent+controller) SSID of the main VAP are change to guest
SSID. It looks like a wrong configuration was pushed by the controller
that changed those SSIDs.
References: PPM-2673
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz> [facelift]
Disable the enable/disable check of wireless interface because of PPM-2672
Keeping the enable/disable command it sould not impact the test suite.
References: PPM-2672
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz> [facelift]
* /010-network.t:
Adapting 'ip route' & ip link expected output.
* /020-ubus.t:
Adapting ubus output
* /030-wifi.t:
Adapting iw dev expected output, because of endpoint addistion.
* /100-prplmesh.t:
Adapting expected output since new guest interface addition
and endpoint enhancement.
Also adapat the output of prplmesh script since
0001-prplmesh_utils.sh-handle-any-number-of-radios-in-bml.patch
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
* /010-network.t:
Adapting 'ip route' & ip link expected output.
* /030-wifi.t:
Adapting iw dev expected output, since new guest interface on 6ghz
had been added and dummy vendor module removed.
* /100-prplmesh.t:
Adapting expected output since new guest interface addition
and endpoint enhancement.
Also adapat the output of prplmesh script since
0001-prplmesh_utils.sh-handle-any-number-of-radios-in-bml.patch
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
* /010-network.t:
Adapting 'ip route' & ip link expected output.
* /030-wifi.t:
Adapting iw dev expected output, since new guest interface on 6ghz
had been added.
* /100-prplmesh.t:
Adapting expected output since new guest interface addition
and endpoint enhancement.
Also adapat the output of prplmesh script since
0001-prplmesh_utils.sh-handle-any-number-of-radios-in-bml.patch
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
* /010-network.t:
Adapting 'ip route' & ip link expected output.
* /030-wifi.t:
Adapting iw dev expected output, since new guest interface on 6ghz
had been added.
* /100-prplmesh.t:
Adapting expected output since new guest interface addition
and endpoint enhancement.
Also adapat the output of prplmesh script since
0001-prplmesh_utils.sh-handle-any-number-of-radios-in-bml.patch
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
upstep pwhm to 4.20.1, libswla to 3.8.0 and libswlc to 5.10.4.
Those version contain deliverie for PPM-2547
close PPM-2547
Signed-off-by: Marouan MCHARFI <marouan.mcharfi_ext@softathome.com>
The move command inadvertently included a backslash, resulting in the
shell treating the asterisk as a literal character rather than a
wildcard. This error in the globbing pattern led to the exclusion of
crucial debug-information in the CDRouter job artifacts. This commit
resolves the issue by removing the backslash, ensuring correct pattern
matching and the inclusion of relevant file.
Fixes: 6cf5635bf50e ("ci: generate coredumps during development")
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently we're trying to have coredumps disabled to avoid regressions
such as PCF-663, but sometimes it might be handy to be able to collect
coredumps from the CI jobs, so lets introduce CI_BUILD_ENABLE_COREDUMPS
variable and enable it during development only, that is not on tagged
builds and not on builds in designated branches.
References: PCF-992, PCF-967
Signed-off-by: Petr Štetiar <ynezz@true.cz>
debug-information is now able to collect coredumps so adjust the jobs
and if available, make those coredumps present in artifacts so they can
be downloaded and examined.
References: PCF-992, PCF-967
Signed-off-by: Petr Štetiar <ynezz@true.cz>
* debug-information: getDebugInformation: collect system core dumps as a fallback
If there is no `FaultMonitor.CoreDumpsDirectory` for whatever reason,
then allow for collecting core dumps defined in `kernel.core_pattern` as
a fallback mechanism.
That sed-fu replaces `core_pattern` specifiers defined in core(5) man
page as following:
* %d,%e,%E,%h,%p to '*' glob pattern
* %c,%g,%i,%I,%p,%P,%s,%t,%u to [0-9]* glob pattern
and doesn't process pipe patterns. So following OpenWrt default core_pattern
kernel.core_pattern = /tmp/%e.%t.%p.%s.core
is going to be converted to:
/tmp/*.*.[0-9]*.[0-9]*.core
shell globbing pattern.
References: PCF-992, PCF-967
* Revert "board-config-haze: add new package with board specific static configuration"
This reverts commit 807633569f9bd1a5462a7294c3dbd6070f4d2e7b as the
networklayout.json generation issues should be fixed in odl-generator in
commit df01b276403b ("netlayout_from_uci.sh: fix sometimes missing
wireless phys config"), delivered in odl-generator v1.6.2 which we're
already using, so there is no need for this workaround anymore.
Fixes: PCF-984
Signed-off-by: Petr Štetiar <ynezz@true.cz>
This reverts commit 04b4955eacf89c38d0068a023b0fcc9ab29c43f5 as it
seems, that the bugs related to the security profile are less prevalent.
References: PCF-967
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Add debug package dependencies for intel_x86 platform to build WLAN debug model for
MxL platforms. On MxL platforms enable debug packages when WLAN debug model is enabled
otherwise enable release packages.
Signed-off-by: Bhavesh Bharadiya <bnbharadiya@maxlinear.com>
After Uninstall() of the DU there is still directory structure present
on the storage in /usr/share/rlyeh/images/prpl-foundation and it seems
like the directory tree is not removed recursively properly for some
reason:
ubus-cli: > SoftwareModules.DeploymentUnit.cpe-c879945e-d002-5775-88a8-e29bc0c641b4.Uninstall()
...
rlyeh: rlyeh_f - [i]removing image 'c879945e-d002-5775-88a8-e29bc0c641b4' [prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64:prplos-v2] - (remove_image@rly
rlyeh: rlyeh_r - [i]Removing index /usr/share/rlyeh/images/prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64/index.json - (rlyeh_remove@librlyeh-v1.1.2/src
rlyeh: rlyeh_r - [i]Removing oci-layout /usr/share/rlyeh/images/prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64/oci-layout - (remove_oci_layout@librlyeh-
rlyeh: rlyeh_r - [i]Removing /usr/share/rlyeh/images/prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64 - (rm_image_dir@librlyeh-v1.1.2/src/rlyeh_remove.c:6
rlyeh: rlyeh_r - [i]Removing /usr/share/rlyeh/images/prpl-foundation/prplos/prplos/prplos - (rm_image_dir@librlyeh-v1.1.2/src/rlyeh_remove.c:655)
thus it makes the CI test fail, as the `/usr/share/rlyeh/images` directory is not empty in the end.
Workaround this by disabling this test temporarily until its fixed.
References: LCM-748
Signed-off-by: Petr Štetiar <ynezz@true.cz>
LCM isn't handling correctly Update() when its being done to a different
location and keeps on pointing the previous image blobs when updating to
a different path, when using two different container images versions but
stored on different locations:
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v1:prplos
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v2:prplos
rather than:
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64:prplos-v1
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64:prplos-v2
Where the version is `v1` and `v2` and not `prplos` for both containers
stored in two different subpaths.
lxc[3860]: BundleVersion: prplos
lxc[3860]: Bundle: prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v2
lxc[3860]: DiskLocation: prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v1
So lets workaround it by providing container in the same path, but tagged with
different versions until the issue is fixed.
References: LCM-747
Suggested-by: Nabil Bizid <nabil.bizid@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
LCM v1.4 introduced network support and network isolation so containers
were adjusted and thus start using them.
References: LCM-737
Signed-off-by: Petr Štetiar <ynezz@true.cz>
LCM v1.4 introduced network support and network isolation so containers
were adjusted and thus start using them.
References: LCM-737
Signed-off-by: Petr Štetiar <ynezz@true.cz>
LCM v1.4 introduced network support and network isolation so containers
were adjusted and thus start using them.
References: LCM-737
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Nabil has pointed out during the review, that this test is no longer
correct. It will always succeed since the verified path doesn't
correspond to the installed image thus it never exist. The verification
should be done on /usr/share/rlyeh/images/prpl-foundation.
Suggested-by: Nabil Bizid <nabil.bizid@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
LCM components were updated to the LCM v1.4 and thus the tests needs to
be adjusted.
Signed-off-by: Matthias FRANCK <matthias.franck_ext@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz> [facelift]
This reverts commit a7f4879cffe5cbb585088fe35b92e2f92a88f767 as this was
fixed with the last LCM update.
Signed-off-by: Matthias FRANCK <matthias.franck_ext@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz> [facelift]
This version includes following updates in the M2/2023 delivery.
* lcm: add liblcm
- The module liblcm provides common functions and datastructures for LCM components
* lcm: bump packages
- cthulhu: bump to v3.5.0
- rlyeh: bump to v2.3.1
- timingila: bump to v2.0.2
- libcthulhu: bump to v0.25.1
- librlyeh: bump to v1.1.2
- cthulhu-lxc: bump to v0.9.1
- timingila-cthulhu: bump to v2.0.2
- timingila-rlyeh: bump to v1.0.3
- cthulhu-dhcpc: bump to v1.2.2
- cthulhu-networking: bump to v0.1.2
* netmodel: bump packages
- libnetmodel: bump to v1.5.14
- netmodel-dhcpv4: bump to v1.3.0
- netmodel-dhcpv6: bump to v1.3.0
- netmodel-ip: bump to v1.4.0
- netmodel-iprouter: bump to v0.3.0
- netmodel-clients: bump to v0.2.0
- netmodel: bump to v2.15.1
* network: add tr181-led
- TR-181 compatible plugin for device LEDs management
* network: add tr181-button
- Implementation of the Button HL API to expose the buttons and its states in (tr181) data model.
* network: add libdiscoping
- Library discovering when IPs appear and disappear
* network: bump packages
- libnetlink-utils: bump to v0.0.9
- mod-dhcpv4c: bump to v0.13.4
- mod-dns-unbound: bump to v1.4.2
- mod-logical-api: bump to v0.6.0
- mod-ppp-daemon: bump to v0.5.5
- mod-vlan-ioctl: bump to v0.1.6
- dhcpv4-manager: bump to v2.3.1
- dhcpv6s-manager: bump to v0.15.2
- ethernet-manager: bump to v1.13.0
- ip-manager: bump to v1.41.3
- netdev-plugin: bump to v1.11.1
- routing-manager: bump to v1.16.0
- tr181-bridging: bump to v2.3.0
- tr181-dhcpv4client: bump to v1.24.1
- tr181-dhcpv6client: bump to v1.1.2
- tr181-dns: bump to v2.3.3
- tr181-dslite: bump to v0.8.2
- tr181-logical: bump to v0.6.1
- tr181-neighbordiscovery: bump to v0.8.0
- tr181-ppp: bump to v0.13.3
- tr181-routeradvertisement: bump to v0.14.4
- tr181-xpon: bump to v1.3.7
- wan-manager: bump to v0.30.0
- libqosmodule: bump to v1.0.1
- mod-fw-amx: bump to v0.3.2
- packet-interception: bump to v0.0.12
- tr181-firewall: bump to v6.9.0
- tr181-pcp: bump to v0.7.0
- tr181-qos: bump to v3.3.11
- mod-dmproxy: bump to v1.3.5
- mod-httpaccess-lighttpd: bump to v0.1.7
- mod-sahtrace: bump to v1.1.0
- deviceinfo-manager: bump to v2.18.4
- multisettings: bump to v0.9.0
- pcm-manager: bump to v0.16.0
- pcm-manager: fix build failure caused by SAH_AMX_PCM_TARBALL
- time-manager: bump to v2.4.2
- tr181-device: bump to v0.15.2
- tr181-dnssd: bump to v0.4.10
- tr181-dynamicdns: bump to master
- tr181-httpaccess: bump to v0.9.0
- tr181-ipdiagnostics: bump to v0.6.0
- tr181-security: bump to v0.3.0
- tr181-upnp: bump to v0.6.1
- tr181-usermanagement: bump to v0.11.0
- hosts-manager: bump to v0.3.1
- libimtp: bump to v2.0.1
- debug-information: bump to v0.1.16
- odl-generator: bump to v1.6.2
* gmap: bump packages
- gmap-client: bump to v1.1.2
- gmap-mod-ethernet-dev: bump to v1.15.1
- gmap-mod-self: bump to v1.4.1
- gmap-server: bump to v2.2.16
- gmap-mibs-common: bump to v0.11.2
- libgmap-client: bump to master
- libgmap-ext: bump to v0.8.0
* ambiorix: bump packages
- acl-manager: bump to v0.6.4
- amx-cli: bump to v0.3.5
- amx-fcgi: bump to v1.3.5
- amxb-inspect: bump to v1.2.1
- amxo-cg: bump to v1.2.14
- amxrt: bump to v2.1.1
- libamxa: bump to v0.10.24
- libamxb: bump to v4.7.16
- libamxc: bump to v1.9.0
- libamxd: bump to v6.0.11
- libamxj: bump to v0.3.62
- libamxm: bump to v0.0.61
- libamxo: bump to v4.2.1
- libamxp: bump to v1.2.12
- libamxrt: bump to v0.2.24
- libamxs: bump to v0.4.3
- libamxt: bump to v0.2.21
- mod-amxb-ubus: bump to v3.1.22
- mod-ba-cli: bump to v0.8.19
- mod-dm-cli: bump to v0.2.5
- mod-lua-amx: bump to v0.1.10
- lua-amx: bump to v0.3.15
Signed-off-by: Matthias FRANCK <matthias.franck_ext@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz> [facelift, PCF-1034 fix]
getDebugInformation script collects all the useful information from the
running system already, so lets use it instead of current list of
commands. While at it, dump the output directly to the serial console
and thus bypass labgrid as it might influence the behavior and/or the
output.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently there is 30 second timeout hardcoded in shell run() method,
but we sometimes might need to use different value, so lets extend the
method args to allow overriding it.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently pcm-manager package fails to build when SAH_AMX_PCM_TARBALL is
enabled:
fs_utils.c:86:10: fatal error: libtar.h: No such file or directory
86 | #include <libtar.h>
| ^~~~~~~~~~
as there is no `libtar` package available, but we didn't noticed this on
the CI as the SDK preserves build root's configuration, so this results
in Config-build.in with following config:
config SAH_AMX_PCM_TARBALL
bool
default n
Which is then default and if the package later changes this to `default
y` like in this case, it has no effect during the built within the SDK.
So lets prevent similar situation in the future and simply remove all
SAH boolean config options from the SDK's Config-build.in and use the
package defaults.
References: PCF-1034
Signed-off-by: Petr Štetiar <ynezz@true.cz>
LCM isn't handling correctly Update() when its being done to a different
location and keeps on pointing the previous image blobs when updating to
a different path, when using two different container images versions but
stored on different locations:
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v1:ynezz-lcm-containers
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v2:ynezz-lcm-containers
rather than:
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64:v1
prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64:v2
Where the version is v1 and v2 and not ynezz-lcm-containers for both
containers stored in two different subpaths.
lxc[3860]: BundleVersion: ynezz-lcm-containers
lxc[3860]: Bundle: prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v2
lxc[3860]: DiskLocation: prpl-foundation/prplos/prplos/prplos/lcm-test-x86-64-v1
So lets fix it by providing container in same path, but tagged with
different versions.
References: LCM-747
Suggested-by: Nabil Bizid <nabil.bizid@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Container rootfs contains the file /etc/resolv.conf which is a symbolic
link to /tmp/resolv.conf thus LCM agent is not able to set up the
network config correctly as it cannot share the DNS settings with the
container through sharing the file /etc/resolv.conf.
lxc-start[1]: <ctr-uuid>: utils - ../src/lxc/utils.c:open_without_symlink:1084 - Symbolic link loop - resolv.conf in /usr/lib/lxc/rootfs/etc/resolv.conf was a symbolic link!
lxc-start[1]: <ctr-uuid>: conf - ../src/lxc/conf.c:mount_entry:2439 - Symbolic link loop - Failed to mount "/lcm/cthulhu/sb_host_data/<ctr>/rootfs/etc/resolv.conf" on "/usr/lib/lxc/rootfs/etc/resolv.conf"
The container can be started without any problem if the container is not
requesting network access or if the file /etc/resolv.conf is either
removed from the container rootfs or at least set a regular file (not a
symlink) which can then be modified at container startup.
So lets workaround that issue by making /etc/resolv.conf regular empty
file.
References: LCM-737, LCM-746
Suggested-by: Nabil Bizid <nabil.bizid@softathome.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently its not possible to customize LCM SDK for our LCM container CI
QA purposes, so lets provide ImageBuilder based LCM testing containers
which can be used for the same purposes.
References: LCM-737, LCM-745, LCM-741
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently its not possible to customize LCM SDK for our LCM container CI
QA purposes, so lets provide ImageBuilder containers which can be used
for the same purposes.
References: LCM-737, LCM-745, LCM-741
Signed-off-by: Petr Štetiar <ynezz@true.cz>
We're going to provide ImageBuilder containers as well, so lets extend
current SDK stage with ImageBuilder.
References: LCM-737
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently we’ve two TLS providers included in the images, mbedTLS and
OpenSSL. This is unfortunate at least from the maintenance, security
and support perspective. The preference is to use OpenSSL as the only
TLS provider, so lets adjust the configuration accordingly.
Fixes: PCF-1006
Signed-off-by: Petr Štetiar <ynezz@true.cz>
