On 6/18/20 5:00 PM, Thomas Lamprecht wrote:
Am 6/18/20 um 4:36 PM schrieb Stefan Reiter:
Legacy IGD passthrough requires address 00:1f.0 to not be assigned to
anything on QEMU startup (currently it's assigned to bridge pci.2).
Changing this in general would break live-migration, so introduce a n
A patch suggestion for the issue described below.
The interesting question is:
why are those "beyond end of device" blocks not zeroed out anyways,
and where does their content come from, given that we have seen
content that does NOT exist in the original image, it appears to be host
memory "garb
>> In second step, we could also add dhcp server features, with static ip/mac
>> leases. (Kea dhcp seem a good candidate).
>> with 1 local dhcp server by node. (only responding to local vms)
>> for bgp-evpn it's easy because we already have a anycast gateway ip, so it
>> can be use by dhcp serv
---
PVE/NodeConfig.pm | 12 +++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/PVE/NodeConfig.pm b/PVE/NodeConfig.pm
index ad49e288..017f6b30 100644
--- a/PVE/NodeConfig.pm
+++ b/PVE/NodeConfig.pm
@@ -25,6 +25,16 @@ PVE::JSONSchema::register_format('pve-acme-domain', sub {
Hi,
currently ACME aliases are restricted to host names and not domain names
(difference explained in [1]),
which prevents the usage of leading underscores in any DNS label.
These aliases are mostly CNAMEs, which allow the usage leading underscores and
should therefore be allowed.
This patch ad
with R-B/T-B and the following follow-up:
[PATCH qemu-server] gen_rand_chars: handle errors properly
should not really happen on modern systems, but random_bytes just
returns false if it fails to generate random bytes, in which case we
want to die instead of returning an empty 'random' string.
S
Look good to me
I tested it and it works. There are no upgrade problems.
Even if jobs already exist.
Regards
Wolfgang
> On 06/17/2020 6:44 AM Wolfgang Link wrote:
>
>
> Hi,
>
> thank you for this patch and the work.
> I will look at this patch and give you feedback.
>
> Regards
> Wolfgang
instead of unconditionally pushing to the 'mounts' array we need to check
if we already have the option in there. without this, we get config
options like:
features: nfs;nfs;nfs
Signed-off-by: Oguz Bektas
---
www/manager6/lxc/FeaturesEdit.js | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
since it prevents boot with our current way of building ZFS modules in
case a system is booted with secureboot enabled.
Signed-off-by: Fabian Grünbichler
---
Notes:
requires an ABI bump
debian/rules | 3 +++
1 file changed, 3 insertions(+)
diff --git a/debian/rules b/debian/rules
index f5
does not fix the issue (which is that the state is not properly resetted
between editing one container and then another one).
On June 22, 2020 1:54 pm, Oguz Bektas wrote:
> instead of unconditionally pushing to the 'mounts' array we need to check
> if we already have the option in there. without
Suggested-by: Lars Ellenberg
Signed-off-by: Stefan Reiter
---
vma-writer.c | 2 ++
vma.c| 2 ++
2 files changed, 4 insertions(+)
diff --git a/vma-writer.c b/vma-writer.c
index fe86b18a60..06cbc02b1e 100644
--- a/vma-writer.c
+++ b/vma-writer.c
@@ -767,5 +767,7 @@ void vma_writer_destroy
Zero out clusters after the end of the device, this makes restore handle
it correctly (even if it may try to write those zeros, it won't fail and
just ignore the out-of-bounds write to disk).
For not even 4k-aligned disks, there is a potential buffer overrun in
the memcpy (since always 4k are copi
initializing 'mounts' array in the panel scope causes edits on subsequent
containers to get the values (mount=nfs) from the previous container. fix this
by
initializing the 'mounts' array in 'onGetValues' and 'setValues'
separately.
Signed-off-by: Oguz Bektas
---
www/manager6/lxc/FeaturesEdit.j
Move the logic which volumes are included in the backup job to its own
method and adapt the VZDump code accordingly. This makes it possible to
develop other features around backup jobs.
Signed-off-by: Aaron Lauterer
---
v6 -> v7: incorporate suggestions, mainly
* change key `data` to `volume_con
as `data` was a bit too generic we now use `volume_config` in the actual
implementations. Thus we should adapt the description as well.
Tab spacing for the other keys has been adapted for easier readabilty.
Signed-off-by: Aaron Lauterer
---
turns out that the `+` in the git diff pushes the desc
The first part of this series which touched pve-manager has been applied
with v6 [0][1].
Missing are the qemu-server and pve-container patches. With the last
suggestions [2] incorporated it is also necessary to update the
AbstractConfig.pm to keep the description of get_backup_volumes() in
sync wi
Move the logic which mountpoints are included in the backup job to its
own method and adapt the VZDump code accordingly. This makes it possible
to develop other features around backup jobs.
Signed-off-by: Aaron Lauterer
---
v6 -> v7:
* change key `data` to `volume_config` in return hash from
ge
Signed-off-by: Aaron Lauterer
---
nothing has changed to previous series
src/PVE/LXC/Config.pm | 24 +---
1 file changed, 17 insertions(+), 7 deletions(-)
diff --git a/src/PVE/LXC/Config.pm b/src/PVE/LXC/Config.pm
index 8d1854a..9fab175 100644
--- a/src/PVE/LXC/Config.pm
++
> I'm just dreaming of something easy like for containers ip management :)
>
>
>
> >>Also, I like the idea that IPAM has a plugin architecture. So it is up to
> >>the plugin to
> >>provide a dhcp service?
>
> I'm still unsure about this. The ipam softwares really only manage subnets,
> and
19 matches
Mail list logo
