- 28 Jun, 2016 40 commits
-
-
Andrew Vagin authored
It uses --manage-cgroup full and can dump and restore cgroup of other tests. Reported-by: Mr Jenkins Cc: Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Acked-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Mike Rapoport authored
Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Mike Rapoport authored
Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Mike Rapoport authored
The deduplication code uses pagemap internal implementation details, so it's better to have this code in the pagemap.c Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Mike Rapoport authored
Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
Here is a race when someone umounted something and this operation isn't propagated into our namespace. CRIU | Another process ----------------------------------------------------------------- pivot_root(".", put_root) | mount(put_root, MS_REC|MS_PRIVATE) | | umount /xxx/yyy | umount /xxx -> EBUSY umount(put_root) We do this to not affect mounts in put_root, but we can mask these mounts as slave and this will work for us and for external users. Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
pids.current is the current number of pids, not the maximum number allowed (which is the thing we care about, that is set by the user). Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
This property can't be written after there are children of the current cgroup or once it has tasks, similar to the cpuset properties. So, let's hoist this property into the "special" array. Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
memory.oom_control when read displays something like: root@kernel:/sys/fs/cgroup/memory/zdtmtst# cat memory.oom_control oom_kill_disable 0 under_oom 0 but when setting it, it expects a "1" or "0" when written. Let's parse out this 1 or 0. Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
Before, as long as we successfully closed the prop fd, the property write would succeed. Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
CC: Andrey Vagin <avagin@openvz.org> Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Mike Rapoport authored
Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Acked-by:
Cyrill Gorcunov <gorcunov@openvz.org> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Adrian Reber authored
Signed-off-by:
Adrian Reber <areber@redhat.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Adrian Reber authored
For the previously added option to skip in-flight connections this adds that option to the RPC interface. The skip in-flight connections is also described in criu.txt. Signed-off-by:
Adrian Reber <areber@redhat.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Pavel Emelyanov authored
This was 1.2 when we merged all file-locks images together. I think it is good time to drop the legacy format. Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Pavel Emelyanov authored
Since 0.3 fdinfo image is ID-s based, so old ones are not used at all. Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Pavel Emelyanov authored
These are raw vaddr:page pairs, criu stopped generating this long before 1.0, nobody uses it now. Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Dmitry Safonov authored
Looks like it's the most appropriate place for those hooks. Didn't know about those helpers - maybe I should have used them in my code more widely instead of casts. Signed-off-by:
Dmitry Safonov <dsafonov@virtuozzo.com> Acked-by:
Cyrill Gorcunov <gorcunov@openvz.org> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Cyrill Gorcunov authored
We will need it for NFS. While we're saving it into image, the test for mode change must be done in account with NFS. So put fixme there. CC: Pavel Emelyanov <xemul@virtuozzo.com> CC: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com> Signed-off-by:
Cyrill Gorcunov <gorcunov@openvz.org> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Cyrill Gorcunov authored
These entries will be used for NFS support and we already test partial support in vz7 version of criu. So to make images compatible lets reserve some of them in advance. CC: Pavel Emelyanov <xemul@virtuozzo.com> CC: Stanislav Kinsburskiy <skinsbursky@virtuozzo.com> Signed-off-by:
Cyrill Gorcunov <gorcunov@openvz.org> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Nikita Spiridonov authored
Needed for container migration, where arguments are set via p.haul as rpc request. Signed-off-by:
Nikita Spiridonov <nspiridonov@virtuozzo.com> Signed-off-by:
Cyrill Gorcunov <gorcunov@openvz.org> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Pavel Tikhomirov authored
1. First save/check all,default,lo together and only after - set them, as setting some sysctls can affect the others 2. Set "accept_redirects" after "forwarding" both for ipv4 and ipv6 3. Change order to "all", "default", "lo" *One can not set ipv6.conf.all.stable_secret, so don't even try, see: commit 622c81d57b39 ("ipv6: generation of stable privacy addresses for link-local and autoconf") Signed-off-by:
Pavel Tikhomirov <ptikhomirov@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Pavel Tikhomirov authored
Restore all/accept_redirects and default/forwarding after all/forwarding as the last can influence the former two. (see inet_forward_change in kernel) # sysctl -w net.ipv4.conf.all.forwarding=1 net.ipv4.conf.all.forwarding = 1 # sysctl -w net.ipv4.conf.default.forwarding=1 net.ipv4.conf.default.forwarding = 1 # sysctl -w net.ipv4.conf.all.forwarding=0 net.ipv4.conf.all.forwarding = 0 # sysctl net.ipv4.conf.default.forwarding net.ipv4.conf.default.forwarding = 0 Trigered with netns-dev test in VZ7CT with vzlinux-6-x86_64 template v2:quirk accept_redirects instead of duplicate entries in devconfs Signed-off-by:
Pavel Tikhomirov <ptikhomirov@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Kirill Tkhai authored
unix: Do not autobind unnamed unix sockets addlen equal to sizeof(addr.sun_family) leads to autobinding of socket. If it had been autobinded when we dumped it, name.len wouldn't have been zero. So, this binding is wrong. Fix that. v2: Merge two conditional branches together Signed-off-by:
Kirill Tkhai <ktkhai@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
Empty cgroups will not be preserved in the criu cgset, since we don't ever dump the criu cgset. This is perhaps a corner case that we may want to address, but is probably (?) really only relevant for test cases. Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
An oddity (bug?) means that we don't preserve empty cgroups in in the criu cgset, since we don't dump it. So we do need to move things into tasks. (This probably got commented out due to a bug in the Ubuntu kernel that I ran across [1] which should be fixed shortly :) [1]: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1588056Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Tycho Andersen authored
Because we're doing this on the same host, the cgroups will already exist. We need to write the properties in full mode to exercise these cases. Signed-off-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Kirill Tkhai authored
Real ino is u32, while parameters are int. Fix that to prevent crap with overflows. Signed-off-by:
Kirill Tkhai <ktkhai@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
CID 163487 (#1 of 1): Unused value (UNUSED_VALUE) assigned_pointer: Assigning value from old_name to pe->name here, but that stored value is overwritten before it can be used. Cc: Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Acked-by:
Tycho Andersen <tycho.andersen@canonical.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Mike Rapoport authored
xemul: There's bfd over pagemap images: mike: Yes, but the bfd caches raw protobuf data, while the in-memory pagemap caches unpacked PagemapEntrie's. There's indeed some duplication, but you anyway have to go through 'bread -> unpack' to use pagemap. Also we read the entire pagemap and then close the image thus freeing the bfd cache. Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Mike Rapoport authored
Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
======================= Run zdtm/static/socket_aio in h ======================== Makefile:295: target 'mnt_enablefs.pid' given more than once in the same rule Start test Makefile:295: target 'mnt_enablefs.pid' given more than once in the same rule ./socket_aio --pidfile=socket_aio.pid --outfile=socket_aio.out Run criu dump =[log]=> dump/zdtm/static/socket_aio/24/1/dump.log ------------------------ grep Error ------------------------ (00.003450) Error (util.c:665): exited, status=3 (00.009104) Error (util.c:665): exited, status=3 (00.024564) Error (util.c:665): exited, status=3 (00.024598) Error (netfilter.c:88): Iptables configuration failed: Success (00.024625) Error (cr-dump.c:1297): Dump files (pid: 24) failed with -1 (00.025117) Error (cr-dump.c:1605): Dumping FAILED. ------------------------ ERROR OVER ------------------------ Wait for zdtm/static/socket_aio to die for 0.100000 Reported-by: Mr Jenkins Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
Now we have many tests with the crfail flags. When criu dump fails, criu may create image files, but doesn't fill them. Reported-by: Mr Jenkins Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
These images are generated by the ip tool. Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Andrew Vagin authored
We can call fchdir into the taget directory and call chroot(".") v2: fix one more comment speaking about proc Signed-off-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-
Adrian Reber authored
Trying to migrate containers with tomcat and running ab as a test client (ab -n 1000000 -c 20 http://url/to/test) criu sometimes exited with an error if in-flight connections were detected. Criu can handle sockets listening and with established connection. In-flight connections can happen on sockets which do not yet have a full established connection (SYN, SYN-ACK, and the last ACK is missing). This adds a new option to criu: --skip-in-flight this option skips in-flight TCP connections. if TCP connections are found which are not yet completely established, criu will ignore these connections in favor of erroring out. With this option criu will skip sockets in this state and let's the client handle the re-connection. Signed-off-by:
Adrian Reber <areber@redhat.com> Acked-by:
Andrew Vagin <avagin@virtuozzo.com> Signed-off-by:
Pavel Emelyanov <xemul@virtuozzo.com>
-