10:37 a.m.
Hi,
i have some questions concerning the destroying of domains, and i hope i'm right here.
If not sorry for the disturbance.
I'm running a two node HA cluster with pacemaker and KVM domains as resources.
From time to time when i try to stop a domain with the cluster manager
that does not work, so the domain is destroyed.
That's ok.
But seldom and irregular also destroy does not work, so the node this domain is running on
is fenced.
That's ugly. Fencing is the worst which can happen to a cluster and i try to avoid
it.
Maybe destroy does not work because of heavy load, i'm currently examing that.
I installed the source package from libvirt-4.0.0, i have a SLES 12 SP4.
I found a virsh-domain.c which i assume is responsible for the domains. Right ?
I found a function called "virDomainDestroy", which i believe has the purpose to
destroy the domain. Right ?
But i don't find the source for that function. I greped the whole source code for it,
but didn't find it.
I'd like to know what this function returns in success or failure.
Could you please help me ?
Thanks.
Bernd
--
Bernd Lentes
Systemadministration
Institute for Metabolism and Cell Death (MCD)
Building 25 - office 122
HelmholtzZentrum München
bernd.lentes(a)helmholtz-muenchen.de
phone: +49 89 3187 1241
phone: +49 89 3187 3827
fax: +49 89 3187 2294
http://www.helmholtz-muenchen.de/mcd
stay healthy
Helmholtz Zentrum München
Helmholtz Zentrum München
11:34 a.m.
On 10/16/20 11:37 AM, Lentes, Bernd wrote:
Hi,
i have some questions concerning the destroying of domains, and i hope i'm right
here. If not sorry for the disturbance.
I'm running a two node HA cluster with pacemaker and KVM domains as resources.
>From time to time when i try to stop a domain with the cluster manager that does not
work, so the domain is destroyed.
That's ok.
But seldom and irregular also destroy does not work, so the node this domain is running
on is fenced.
That's ugly. Fencing is the worst which can happen to a cluster and i try to avoid
it.
Maybe destroy does not work because of heavy load, i'm currently examing that.
I installed the source package from libvirt-4.0.0, i have a SLES 12 SP4.
The first thing to try, if there is any possibility, is to upgrade your
libvirt version. 4.0.0 is nearly 3 years out of date at this time, and
*a lot* has changed since then. If you actually are experiencing a
libvirt bug, then it's very likely it was identified and fixed during
this time. (Even though you are running and older version of SLES, it's
likely someone somewhere has built a newer version and has a pre-built
package available)
I found a virsh-domain.c which i assume is responsible for the
domains. Right ?
virsh-domain.c contains the functions that correspond to domain-related
commands of the virsh utility. They gather input from the commandline,
then call libvirt's public API. So you won't find much there.
I found a function called "virDomainDestroy", which i
believe has the purpose to destroy the domain. Right ?
That is the top-level public API (what is called by user programs like
virsh or virt-manager). It looks into the virConnect object that was
previously opened by said user program, and calls the domainDestroy()
function for whichever hypervisor driver you're using. So for example,
if you are using the QEMU driver, it will end up calling
qemuDomainDestroy(), or if you're using xen (libxl), it will call
libxlDomainDestroy() (you can get a list of all these functions by
searching for "domainDestroy" in the source - each hypervisor will
initialize the .domainDestroy member of its virHypervisorDriver object
with a pointer to its domainDestroy function.)
From there, you can search for the appropriate function for the
hypervisor you're using. For example, qemuDomainDestroy is in
src/qemu/qemu_driver.c, and libxlDomainDestroy is in
src/libxl/libxl_driver.c
But i don't find the source for that function. I greped the whole
source code for it, but didn't find it.
If you grep'ed all the source you have for virDomainDestroy and didn't
find it, then either you don't have all the source, or there was a
mistake in your grep command :-). virDomainDestroy is in
src/libvirt-domain.c (but as I detailed above, that's not actually the
function you're looking for, only a switch-point in the long linkage of
the journey from application to hypervisor driver).
In the end, while I don't want to discourage looking into the source in
general, since it's always enlightening and leads to a better
understanding, this may not be the most efficient path to a solution;
you may get better, faster results by upgrading to as new a version of
libvirt as possible, then if the error continues, come back to the list
with a detailed description of the error (or if you think direct
interaction could lead to a quicker resolution, point an IRC client at
irc.oftc.net, join the #virt channel, and ask away - several libvirt
developers are there during weekday normal worktime (mostly in Europe,
plus a few in the U.S.), and we try our best to be helpful and polite :-)
Good luck!
I'd like to know what this function returns in success or
failure.
Could you please help me ?
Thanks.
Bernd
6:09 a.m.
----- On Oct 16, 2020, at 6:34 PM, Laine Stump laine(a)redhat.com wrote:
I installed the source package from libvirt-4.0.0, i have a SLES 12 SP4.
The first thing to try, if there is any possibility, is to upgrade your
libvirt version. 4.0.0 is nearly 3 years out of date at this time, and
*a lot* has changed since then. If you actually are experiencing a
libvirt bug, then it's very likely it was identified and fixed during
this time. (Even though you are running and older version of SLES, it's
likely someone somewhere has built a newer version and has a pre-built
package available)
Yes, but installing a more recent version not from the official SLES repository makes me
loose
my support.
> I found a virsh-domain.c which i assume is responsible for the
domains. Right ?
virsh-domain.c contains the functions that correspond to domain-related
commands of the virsh utility. They gather input from the commandline,
then call libvirt's public API. So you won't find much there.
This is the part i want to keep an eye on (virsh-domain.c)
:
static bool
cmdDestroy(vshControl *ctl, const vshCmd *cmd)
{
virDomainPtr dom;
bool ret = true;
const char *name;
unsigned int flags = 0;
int result;
if (!(dom = virshCommandOptDomain(ctl, cmd, &name)))
return false;
if (vshCommandOptBool(cmd, "graceful"))
flags |= VIR_DOMAIN_DESTROY_GRACEFUL;
if (flags)
result = virDomainDestroyFlags(dom, VIR_DOMAIN_DESTROY_GRACEFUL);
else
result = virDomainDestroy(dom); <=====================================
if (result == 0) {
vshPrintExtra(ctl, _("Domain %s destroyed\n"), name);
} else {
vshError(ctl, _("Failed to destroy domain %s"), name);
ret = false;
}
virshDomainFree(dom);
return ret;
}
Does that mean that virsh itself returns a true and write "Domain %s
destroyed\n" if it successfully
destroyed the domain and otherwise returns a false and writes "Domain %s
destroyed\n" ?
> I found a function called "virDomainDestroy", which i
believe has the purpose to
> destroy the domain. Right ?
That is the top-level public API (what is called by user programs
like
virsh or virt-manager). It looks into the virConnect object that was
...
If i want to check success or not with virsh then the top-level API is enough for me.
Bernd
Helmholtz Zentrum München
Helmholtz Zentrum Muenchen
Deutsches Forschungszentrum fuer Gesundheit und Umwelt (GmbH)
Ingolstaedter Landstr. 1
85764 Neuherberg
www.helmholtz-muenchen.de
Aufsichtsratsvorsitzende: MinDir.in Prof. Dr. Veronika von Messling
Geschaeftsfuehrung: Prof. Dr. med. Dr. h.c. Matthias Tschoep, Kerstin Guenther
Registergericht: Amtsgericht Muenchen HRB 6466
USt-IdNr: DE 129521671
11:18 a.m.
On 10/16/20 9:37 AM, Lentes, Bernd wrote:
Hi,
i have some questions concerning the destroying of domains, and i hope i'm right
here. If not sorry for the disturbance.
Laine already answered your questions about code navigation. I'll provide a few
random comments.
I'm running a two node HA cluster with pacemaker and KVM domains
as resources.
>From time to time when i try to stop a domain with the cluster manager that does not
work, so the domain is destroyed.
That's ok.
But seldom and irregular also destroy does not work, so the node this domain is running
on is fenced.
Have you investigated why the associated qemu process is not responding to
SIGKILL? Is it blocked on uninterruptible I/O? E.g. are there any hints in
/proc/<qemu-pid>/stack or /proc/<qemu-pid>/wchan?
That's ugly. Fencing is the worst which can happen to a cluster
and i try to avoid it.
Maybe destroy does not work because of heavy load, i'm currently examing that.
I installed the source package from libvirt-4.0.0, i have a SLES 12 SP4.
Laine mentioned trying a new libvirt. While that is possible, it can be
difficult in practice due to missing or insufficient build dependencies. E.g.
SLES12 SP4 does not have meson. Creating a SLE12 SP5 based build container with
all dependencies to build upstream libvirt is on my todo list. ATM I was only
considering doing this for the latest SLE12 service pack.
Regards,
Jim
1493
days inactive
1500
days old
3 comments
3 participants
participants (3)
-
Jim Fehlig -
Laine Stump -
Lentes, Bernd