Released vzctl-3.0.25.2This is a next stable update to 3.0.25, fixing the following bugs:
* vzeventd: do not leave zombies
* vzeventd: fix logging of killed children
* vps-functions: make sure PATH contains dirs we need
* rpm: add /etc/sysconfig/vzeventd as %ghost %config
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
vps-functions: make sure PATH contains dirs we need'arpsend' is installed into our @SBINDIR@ (which can be /usr/sbin
or /usr/local/sbin or something else).
'ip' usually lives in /sbin or in some cases /bin. Previously
(in vzctl <= 3.0.24) we called in /sbin/ip, but commit 8894720
changes that.
So, make sure all these dirs are in PATH, so we don't fail miserably,
for example like this:
$ sudo /usr/sbin/vzifup-post eth1
/usr/lib64/vzctl/scr...
vzctl.spec: add /etc/sysconfig/vzeventd as %ghostTeach RPM that if this file exists then it's ours.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
(cherry picked from commit e816ee05792dbd8038c905f06cb557acae05e840)
vzeventd: fix child handler to avoid zombiesSometimes a signal is not delivered (and child_handler() is not called)
for every single child exited or terminated, which causes zombies to appear.
Use waitpid() and loop until all the children are properly reaped.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
(cherry picked from commit d7ac02f58412526632ec7751aee56c04580be988)
src/vzevent.c: fix logging of child killedCheck for WIFSIGNALED was done under WIFEXITED which is just
plain wrong, therefore we never got "Child killed" message. Fix it.
While at it, change the message from 'was terminated' to just 'killed'.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
(cherry picked from commit 11ff8a43cb14b4b576a2810b74297cc7f419c4b0)
src/vzevent.d: simpler child handlerUse old-style singal handler function, we don't really need any of the
fields provided by struct siginfo_t, and pid is returned by wait().
Pure simplification, should not have any visible effect.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
(cherry picked from commit 8afe85a38f4e4ae3e9bea8d3cfee794a119e5c9d)
Released vzctl-3.0.25.1This is a stable update to 3.0.25, fixing the following regression:
* vzctl set --devices: do not segfault (#1729)
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
vzctl set --devices: do not segfault (v2)This is an addition to commit ea64ee6b9. It fixes segfaults
in four different places when either '--devices' command line
option and/or 'DEVICES=' parameter in configuration file is used.
http://bugzilla.openvz.org/1729
Reported-by: Vyacheslav Nikitin <alien.virus@gmail.com>
Reported-by: Paparaciz <helpaz@gmail.com>
Reported-by: frank <frank@si.ct.upc.edu>
Signed-off-by: Kir Kolyshkin <kir@op...
vzctl set --devices --save: do not segfaultThis is an addition to commit ea64ee6b9. It fixes segfault
in store_dev() when --devices option was used.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
Released vzctl 3.0.25 CHANGES
(since vzctl-3.0.24.2)
* Removed cron scripts to handle cleanup and reboot (#1423)
* Introduced vzeventd to handle cleanup and reboot (#1519)
* vzctl start/runscript: add ability to set per-CT osrelease
(/etc/vz/osrelease.conf) -- support for Fedora 14 etc. (#1709)
* vzctl set --pci_{add,del}: add the ability to move PCI devices to CT
* Removed FAKEGATEWAY routing for all distro...
src/vzeventd.c: fix compile on CentOS 4Old and ancient CentOS4 glibc-kernheaders package provides
/usr/include/linux/netlink.h which does not #include <linux/types.h>
so we have to do it explicitly, otherwise:
CC vzeventd.o
In file included from vzeventd.c:25:
/usr/include/linux/netlink.h:22: error: syntax error before "__u32"
/usr/include/linux/netlink.h:28: error: syntax error before "__u32"
/usr/include/linux/netlink.h:30:...
vzctl: allocate a device name dynamicallyBefore this patch a device name has been limited by 32 symbols
Now we want add input device for exmple:
/dev/input/by-path/pci-0000:00:12.0-usb-0:3:1.1-event
Signed-off-by: Andrey Vagin <avagin@openvz.org>
scripts/vps-create: fix VE_PRVT permissionsScript's $VE_PRVT is something like /vz/private/100.tmp, which will
be renamed into the new container's tree root directory
(/vz/private/100 in this example).
Some templates do not include an entry for "." or "/". When these are
used (with "vzctl create") on a host system with umask 077 (Owl's
default), this resulted in the new system's root directory having
mode 700, and in turn in all sorts ...
scripts/vps-create: use -Sp options for tar-S: handle sparse files more effectively
-p: ignore umask, use exact permissions from the archive
Reported-by: Solar Designer <solar@openwall.com>
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
vzctl set --features bridge: enable CAP_NET_ADMINCapability CAP_NET_ADMIN is required for VE_FEATURE_BRIDGE
to work, so let's raise it automagically.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
etc/dists/scripts/*set_ugid_quota.sh: fix mtab permsIn *set_ugid_quota.sh scripts, set the permissions on /etc/mtab to 644
regardless of the current umask.
Fix enabling sysfs feature by defaultThe problem caused by this bug is if some other features are
explicitly set, SYSFS is not enabled by default.
From Solar Designer:
"""
This commit looks buggy to me:
http://git.openvz.org/?p=vzctl;a=commit;h=vzctl-3.0.23-262-g2660464
- if (!(res->env.features_known & VE_FEATURE_SYSFS) &&
- sysfs_required(res)) {
+ /* sysfs enabled by default, unless explici...
etc/ve-basic.conf-sample: increase LOCKEDPAGESRecent distros (Fedora 14 and Ubuntu 10.10) want more.
Make them happy.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
vzctl start/runscript: add ability to set per-CT osreleaseSome distros (like recent Fedora versions) expect a kernel
version to be x.y.z or newer, and won't work with older kernels.
Our kernels are usually sufficiently patched to work with new
userspace, but this check for kernel version requires some
trickery.
Introduce /etc/vz/osrelease.conf to have these requirements
configurable. Use new /dev/vzctl ioctl to set per-CT osrelease
field, based on di...
scripts/.gitignore: updateRecent commits (like 4cc6257616 and 5b3579775) removed some
scripts, added some different ones. Update gitignore accordingly.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
src/arpsend.c: fix a compiler warningFix a long-standing compiler warning:
arpsend.c: In function ‘create_arp_packet’:
arpsend.c:354:3: warning: missing initializer
arpsend.c:354:3: warning: (near initialization for ‘(anonymous).sndr_hw_addr’)
Should have no visible effect.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
src/lib/exec.c: flush stderr/stdout fefore fork() in vzctl_exec*()Otherwise it might be flushed twice after fork(),
producing extra garbage as a result.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
quota_ctl(), vps_set_quota(): refactor/fixThere is no need to do hairy trickery with return values
in quota_ctl() -- always return the exit code of vzquota.
In vps_set_quota(), when we run quota_ctl(STAT2), we need
to find out if UGID quota is running or not. If it's not
running vzquota returns exit code of 9, which we check for.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
vps_set_quota(): return proper error codeWe should return VZ_DQ_SET here in case of error like this,
because this code is used as an exit code of vzctl and -1
just doesn't make sense here.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
src/lib/quota.c: don't check for dq_param == NULLThere is no need to check for dq_param == NULL here, it can
never be NULL.
Also, the return code of -1 is bad because these functions are
expected to return codes suitable as vzctl exit codes.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
man/vzctl.8: add/describe 'vzquota on failed' exit code (60)After commits d75908d0 and 84f9ab9329 vzctl may now exit with
VZ_DQ_ON (60) exit code, which means 'vzquota on failed'.
Document it.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
quota_on(): refactor and fixThis is more of a fix than proper refactoring.
So what I tried to achieve is:
1. Return proper code (one of VZ_DQ_ON, VZ_DQ_INIT, VZ_DQ_SET)
depending on what really failed.
2. Do not print errorneous "vzquota on failed" with wrong exit code
for the case of failed quota_init() or quota_set(). Note that
quota_init() and quota_set() print their own error messages
and return correct...
vzctl set --devnodes: always try to recreate device in CTIf device file already exists inside a container, do not trust it,
it might be bad (not a device file, invalid/old major/minor etc).
Always re-create device file from scratch.
Reference: PCLIN-28558
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
vzctl quotaoff: fix wrong exit code if quota is not runningIf quota is not running, we should not return an error
when trying to turn it off. There is such a code already,
but it forgot to set ret to zero, therefore the native
exit code from vzquota (6) was returned by vzctl which
makes no sense (for vzctl 6 means "Not enough system
resources").
Fix it by 'ret=0' in case quota is not running.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
vzctl --help: improve/reformat, add quota* commandsFirst, let's just use one single fprintf. Should not have
any visible effect per se.
Second, make sure we fit into 80 columns (moving --netif_del to
another line).
Third, change the order of commands (first create, then start,
then anything else).
Fourth, introduce a few empty lines for better readability.
Finally, add quotaon/quotaoff/quotainit commands.
Signed-off-by: Kir Kolyshkin <kir@...
vzctl quotaon|quotaoff|quotainit: report and return error if disk quota is disabledIf disk quota is disabled (either globally or for this CT) do not just
silently skip this step, but print an error message and return an appropriate
error code.
Signed-off-by: Kir Kolyshkin <kir@openvz.org>
Kir Kolyshkin
