When we configure dhcp interfaces before network.target has run,
network.target will try to bring up those interfaces a second time
after our service does so. This causes two issues - first, the
network target will always fail because it can't bring up an
interface that is already up, and second, when configuring interfaces
that don't actually have an available DHCP server it will result in
a five minute delay waiting for DHCP on those interfaces. This will
also cause the network target to fail and is an unnecessary delay.
By moving the dhcp-interface service to run after the network
target we avoid both of these problems. network.target will still
bring up the interfaces on subsequent boots. This could result in
the five minute delay happening on reboots, but the expected use
case for interfaces without DHCP is that they would be configured
statically on initial deployment so this should be a minor issue.
The dhcp-interface service is also configured to run before the
network-online target so that services which depend on the network
actually being available will not race the DHCP process.
A snippet from /var/log/messages on a node with this patch applied
is included in the bug to demonstrate the behavior described above.
Change-Id: I5cfabf20f920beea52abf4c42362b6f6ac0b37c4
Closes-Bug: 1653812
systemd doesn't like it when service files have the executable bit
so this causes it to spam the journal with messages like:
Configuration file /usr/lib/systemd/system/dhcp-interface@.service is
marked executable. Please remove executable permission bits.
Proceeding anyway.
Removing the executable bit from the install permissions should
eliminate those messages.
Change-Id: Ie1bc39465b3fcb55dcda5cee9e46a128a6ccffcb
DIB_INIT_SYSTEM is exported by the dib-init-system element and contains
the output of the dib-init-system script so there is no need to
re-initialize it during various phases.
Change-Id: I09d6d10742689efe3d8eb9d64b539d6599b46227
On Centos and RHEL 6 the init system is upsart but but networking is using
sysv compatabiliy and a code path the handle this situation.
We can't use DISTRO_NAME because the centos-minimal element sets it to
centos for CentOS 7 but the centos element sets it to centos for CentOS 6.
Change-Id: Ib8e33ed78b3d6a5737eb7449bccef2d33f72b131
Closes-Bug: #1638527
We are currently wasting about 10 minutes per deploy waiting for
DHCP on interfaces that will never get it. By default, the timeout
seems to be 5 minutes (the 10 minutes is because we boot both the
IPA ramdisk and the deployed image, and each waits for 5 minutes),
which is excessively long to get a DHCP response. This change
shortens the time to 30 seconds. If an interface hasn't gotten a
response in 30 seconds, chances are it's not going to. A 30
second wait should reduce our wasted time to 1 minute, which is
more reasonable.
This is being done in the systemd unit file because the -timeout
option to dhclient doesn't seem to override what is configured in
dhclient.conf, and doing it in the systemd file means that this
change will be limited to only the interfaces configured by
dhcp-all-interfaces.
Change-Id: Ia8610e3def39c937eb0c861fdc9bc571ec39f9f4
Closes-Bug: 1626673
On systemd-based operating systems that don't
use /etc/sysconfig/network-scripts
dhcp-all-interfaces configures 'lo' for dhcp.
This causes errors and fails networking.target
causing system-wide issues. This change excludes
'lo' at dhcp-all-interfaces udev rules level.
Closes-bug: #1621501
Change-Id: I7563b766827bedbea7ae1de35e5bdfcbf1fc0d1e
Co-Authored-By: Jeremy Stanley <fungi@yuggoth.org>
Dependency to start network-pre (which
depends on network.target) before
dhcp-interface@.service collides with
Ubuntu's own network.target that suupose
to start after network-pre.
Change-Id: I9e59c970bfb1ebdaa15b4ec6b545761ede3ca056
Closes-bug: #1619816
Currently there is no way for a service to become aware that
dhcp-all-interfaces is finished configuring all the interfaces at
boot time. This causes problems for applications like the
ironic-python-agent which scans the interfaces when it first starts as
part of the inspection stage and can race against dhcp-all-interfaces
bringing up the interfaces, leading to inconsistent results.
This patch ensures that the dhcp-all-interfaces script runs before any
network interface is configured and brought up by the rest of the
system, and also ensures that the ironic-agent element also waits for
the network to be online before starting. This is done by using the
network targets provided by systemd.
Change-Id: Id9583b7f54361aa603a6229da598ad6a0f0f7938
The dhcp-all-interfaces and simple-init elements did not have the ISC
DHCP Client package mapped for OpenSUSE, which caused DIB to fail with
"'isc-dhcp-client' not found in package names. Trying capabilities."
Similarly, the bootloader element did not have the grub-pc package
properly mapped for OpenSuse, which caused DIB to fail with "Package
'grub-pc' not found.".
This patch adds the package mappings for these elements so that the
opensuse element can be created and booted successfully.
Change-Id: Ife478158fec3a95de73a9206b38dcc6511d56cc8
Every platform has a different name for their DHCP client, so use a
generic name "dhcp-client" in the package name and let everyone choose
their sub-name. This also brings some consistency across simple-init
& dhcp-all-interfaces
Change-Id: I797aa7aacb13dfb7f35700463dc11d55552eb108
This makes use of the dhcpcd package and it's ability to run on all
interfaces by default. We disable the privacy extensions and dhcp
overriding the hostname (both are enabled by default). Other than
that it 'just works' and was the method used to bring up interfaces
on Gentoo Openstack images before we switched to building with DIB.
Change-Id: I02c14927d70b22f560c6fc149fefca0f93933f56
On ubuntu, if an interface exists but is unconfigured then ifquery
$iface will return 0 but no output. This will cause dhcp-all-interfaces
to not attempt to dhcp on that interface.
Change-Id: Ic1532728ae9ca6d759714392f727d25b814ef336
As the doc says, we want to remove all interfaces that come by default
in cloud images. Cloud images also come with an eth1 though, so lets
just remove all interfaces.
Change-Id: I49c74a3285e6a610ac723bdf976f3727d2736749
InfiniBand interface takes more time to bring up then
Ethernet interface. This patch just increase the retries
to 20 times, to make it work for InfiniBand as well.
Change-Id: I5c4842696207885552413ea2d053f2e90bd6803c
Always no quotes for $() statement.
We don't need quotes to hold blanks in result:
# i=$(echo 1 2 3)
# echo $i
1 2 3
#
These quotes can make something wrong in some case:
# i=$(echo '!')
#
# i="$(echo '!')"
-bash: !: event not found
#
No real problem for current code, only to use a better code style.
Change-Id: I5909636bdc8de3d44a305d033c8c892af446acf3
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
There is a wide variety of tracing options through the various shell
scripts. Some use "set -eux", others explicity set xtrace and others
do nothing. There is a "-x" option to bin/disk-image-create but it
doesn't flow down to the many scripts it calls.
This adds a global integer variable set by disk-image-create
DIB_DEBUG_TRACE. All scripts have a stanza added to detect this and
turn on tracing. Any other tracing methods are rolled into this. So
the standard header is
---
if [ "${DIB_DEBUG_TRACE:-0}" -gt 0 ]; then
set -x
fi
set -eu
set -o pipefail
---
Multiple -x options can be specified to dib-create-image, which
increases the value of DIB_DEBUG_TRACE. If script authors feel their
script should only trace at higher levels, they should modify the
"-gt" value. If they feel it should trace by default, they can modify
the default value also.
Changes to pachset 16 : scripts which currently trace themselves by
default have retained this behaviour with DIB_DEBUG_TRACE defaulting
to "1". This was done by running [1] on patch set 15. See the thread
beginning at [2]
dib-lint is also updated to look for the variable being matched.
[1] https://gist.github.com/ianw/71bbda9e6acc74ccd0fd
[2] http://lists.openstack.org/pipermail/openstack-dev/2014-November/051575.html
Change-Id: I6c5a962260741dcf6f89da9a33b96372a719b7b0
We are autodetecting and configuring devices in dhcp-all-interfaces
so having one configured in /etc/network/interfaces by default
is redundant and slows boot down.
Change-Id: Ic4e8a0668c793d21ed2dd96908649c9a77264f67
Closes-Bug: #1239480
I don't actually think there is a way that disable_interface
would get called on Fedora... because if config_exists
it takes a different path. Furthermore we really don't
want to mess w/ configured interfaces in dhcp-all-interfaces
because that could be side effecty (and is arguably the
job os os-net-config anyways).
This function already did nothing on Debian so removing
shouldn't hurt anything.
Change-Id: Ic53bbab98aa03d487efd290ae8e076045781b5b1
As with the previous similar changes, this is intended to catch
problems as they happen instead of ignoring them and continuing on
to potentially fail later. Setting this on all existing scripts
will allow us to enforce use via Jenkins.
Change-Id: Iad2d490c86dceab148ea9ab08f457c49a5d5352e
Using set -e in all of our scripts will prevent some subtle bugs
from slipping in, and will allow us to enforce use of set -e with
tooling.
This change also adds -u and set -o pipefail in the less complex
scripts where it is unlikely to cause problems. A follow-up change
will enable those options in the complex scripts so that if it
breaks something it can be reverted easily.
Change-Id: I0ad358ccb98da7277a0ee2e9ce8fda98438675eb
This patch adds support for sysv init to the dhcp-all-interfaces element.
Specifically, it adds a sysv init script and the relevant install/config steps.
Tested with Debian jessie/testing.
Change-Id: I4ae1d5365bafda188cb65ea1fb72d8ffe2408910
I3166951c6fbb2faec53ffc339edd2fd0c3512300 recently landed without two
fixes that I suggested, namely using rm -f rather than testing for
file existance, and a comment fix.
Change-Id: I68e187e20802f3482d4c524a6c435fec7526178e
Moves the code that removes default cloud interfaces
into the dhcp-all-interfaces element. Having it live in
stable-interface-names won't work right now because we
need this for virtualized testing.
Change-Id: I3166951c6fbb2faec53ffc339edd2fd0c3512300
This was calling it for all interfaces previously, making the fixes
we'd put in place be ineffective.
Change-Id: I16bf81225a942e3b19a7e745d949e6ddea2fd50e
When high VM churn is encountered dhcp-all-interfaces got tangled up
and leaked instances until upstart run out of filedescriptors.
This happened because we were locking on every invocation, and the
lock logic was disabling the single-interface code - so we always
probed every interface every time - and this was serialized across
every instance of dhcp-all-interfaces.
Change-Id: I4ef1fb83d978cc99d4cb9cc2666bde302c6db7d1
Corrects an issue in the config_exists function that could
cause duplicate interfaces to be configured on Ubuntu.
Previously we used 'return ifquery' directly which caused
a silent 'numberic argument required' error to go unnoticed.
This would also return 1 meaning a new interface would get
configured.
The new logic should handle things properly.
Change-Id: I625225e15113d7e184e3bcb5054df1616dec008a
Closes-bug: #1298430
Updates the dhcp-all-interfaces element to fix a race
with the recent udev rules implementation on Fedora.
With the new approach we make the udev rule want (require
to startup) a generic dhcp-interface@.service template which
can be started individually for each interface that is
discovered.
The dhcp-interface@.service is setup such that it:
1) It calls dhcp-all-interfaces <iface> directly with
a pre-exec script. This creates the ifcfg file right
before we need it but avoids the case where network.service
might get greedy and try to start it itself.
2) Only runs if the ifcfg script doesn't already exist. This
is important because we only need to bootstrap the DHCP configs...
Once they exist the network.service will take care of starting them
on reboots, upgrades, etc.
3) On initial boot ensure that the initial DHCP interfaces come
up after network.service. Since we really only want
dhcp-all-interfaces to help bootstrap that haven't already
been configured this seems reasonable.
4) We also try to ensure that cloud-init
comes up after the DHCP interfaces. Cloud init has a decently
long timeout that this wasn't a functional problem but it keeps
log file spew down.
Change-Id: I71b026f027182aad49c3435bb903e5e38e524685
Closes-bug: #1294803
Refactors dhcp-all-interfaces.sh so that if an optional
INTERFACE argument (the first argument) is passed to the script
it only inspects that single interface. If no argument is
passed then the previous default behaviour is used which
causes all interfaces to be inspected.
To avoid a collision with the previous $1 we move to using
$FLOCKED for the exec flock command which runs on distributions
using ENI.
Also sets PATH so that the commands within the script
can all be found if it isn't set properly (/sbin/ip, /bin/cat, etc.)
This is a move towards using udev rules to add these types
of interfaces automatically.
Change-Id: I3ec8fd2cc2071bfc6943c744ca619e31b71146fc
Updates the dhcp-all-interfaces element so that
we add new DHCP interfaces using a udev rule
instead of a systemd service.
The previous approach (using the systemd service)
was proving to be problematic when used with
the new Fedora constistent network naming
convention rules (em1, em2, etc.). We aren't using
these device names as a default yet... but we will
need to soon and when we do udev rules seem to be the
best way to avoid racy behaviour when calling
dhcp-all-interfaces.sh from the systemd service.
Change-Id: I87e0aed1f34718e896bd414388886a1f1073b0c0
Refactors dhcp-all-interfaces.sh so that if an optional
INTERFACE argument (the first argument) is passed to the script
it only inspects that single interface. If no argument is
passed then the previous default behaviour use used which
causes all interfaces to be inspected.
Also sets PATH so that the commands within the script
can all be found if it isn't set properly (/sbin/ip, /bin/cat, etc.)
This is a move towards using udev rules to add these types
of interfaces automatically.
Change-Id: Ia482c1d3ddce0f0d8d77f9bc3ac76d6924640715
Updates dhcp-all-interfaces so that we avoid using
ifquery on Fedora to determine if an interface exists.
Adds a new config_exists function which simply looks for
the ifcfg config file on distro's that use netscripts.
Change-Id: Ie55524b83820fe4fab28cc9d3f7e08ca2e42c182
Rename generate-interfaces-file.sh to dhcp-all-interfaces.sh
so that it is easier to find/remember in the $PATH on a deployed
image.
Also, on some distros the script actually generates many files
so the previous name was a bit confusing.
Change-Id: I7152fa7c28e8ade251311da2cd5f75972423b66c
Updates the generate-interfaces-file.sh script in
the dhcp-all-interfaces element so that we only
add interfaces that have real MAC addresses.
The generate-interfaces-file.sh script is run early enough
in the boot process (before OVS is initialized) that this
isn't usually a problem unless you execute it manually by hand
after booting. Then you'll end up with network/DHCP
configs for all of your OVS bridges, etc. This
patch avoids configuring all of the virtual interfaces
which have generated MAC addresses.
Change-Id: I7a705084aa5b11305ac0ec5ca37fd2e87a2ae8b7
Closes-bug: 1239479
Updates the dhcp-all-interfaces element so that the link
check tries up to 10 times.
On some of my machines 3 times (seconds) doesn't appear to
be quite long enough...
Change-Id: Ibf7015162fc890d4de8a417b868b0301146944c0
Includes:
* Enhancements to the generate-interfaces-file.sh script
so that it generates ifcfg network-scripts for
distros like Fedora/RHEL.
* Includes a new dhcp-all-interfaces.service systemd
script which ensures network interfaces get generated
before the network service starts on Fedora.
* Add a new disable_interface function to
generate-interfaces-file.sh which (on Fedora)
will delete the ifcfg config for an interface which
isn't plugged in.
This is important because ifcfg-eth0 exists by default
and we want to remove it if the NIC isn't connected.
Previous behaviour on Ubuntu is unchanged and the
generate-interfaces-file.sh just logs it as skipped.
* General doc updates...
I tested this on a multi-nic machine w/ Fedora where the first NIC
exists but was not connected.
Change-Id: Ia99e312539da43caefd72aa60398d43dac5dcc8f
Closes-Bug: 1239880
Since we are using bash syntax in some of the element fragments,
we should make sure we use bash for all of them, so that things don't
break on systems where /bin/sh != /bin/bash.
Change-Id: If2f043c57aa4e1492b7f9839213ef6123f683612
If the /etc/init drectory gets created on Fedora, it causes
os-svc-daemon to add upstart init scripts instead of systemd.
As a result none of the openstack services can start.
Partial-Bug: #1239880
Change-Id: Iec317baa3eb9ff651fa66c582d2f614993cde45e
For some reason we are adding CR's to/etc/network/interfaces, but it
does not require them. They are a bit annoying to see in an editor and
serve no known purpose there.
Change-Id: I9aeeff5533f418f09fcf33edd42e5d85cd486d23
