Some database tasks are reporting:
TypeError: not enough arguments for format string
This patch cleans up these log calls
Change-Id: I895ec902cb077df6eb6ae950e0bb1f0b9e40d76f
Closes-Bug: #1513302
Local certificate manager is broken and should probably be removed,
since the effort was never taken to make it functional (adding
additional fields for taking raw certificate data in the API).
Barbican should be the default as it is the functional option.
Also taking this opportunity to clean up the config and local.conf
defaults to match reality.
Change-Id: Icf1e89838b761e7d2926c8eb7532404b4c2f19dd
The field for pool.lb_algorithm is too small to hold the value
"LEAST_CONNECTIONS." This patch expands the size of this field.
Change-Id: I77c7b5136a0427e27897f24ad0a8409818bd5c21
Closes-Bug:1464962
This patch introduces stricter checking on fields that are in
lookup tables, but are effectively Enum fields. Because we do a
lot of case-sensitive string comparisons against these fields in
various places throughout the code tree, but because SQL does not
to case-sensitive comparisons by default, it's important that we
are case-sensitive in the API for these fields.
Change-Id: Ib23796754ca54f69bb60c79c75804a5e7ee26e29
Closes-Bug:1534217
In case when update data for a pool contains session persistence
it is required to update session persistence table first and
then remove 'session_persistence' key from the update dict.
Otherwise, update will fail and a pool will get stuck in
PENDING_UPDATE status.
This change reuses update_pool_on_listener method of
db.repositories.Repositories class in UpdatePoolInDB task instead
of PoolRepository update method.
This regression was discovered using a scenario test:
https://review.openstack.org/#/c/207945/
Change-Id: I53d6b7962c0baa466db7e053157cdd302edae473
Closes-Bug: #1505125
This patch disables the default haproxy process from starting in
the amphora image.
Closes-Bug: #1527691
Change-Id: Iccac03b982e47fc85f8e6beb0cc55a80b73ab465
Get_delete_member_flow original tasks order was:
1. DeleteModelObject
2. ListenerUpdate
3. DeleteMemberInDB
4. MarkLBAndListenerActiveInDB
Since memebrs are deleted from Listener only by DeleteMemberInDB,
ListenerUpdate is being called with all members (including the
member that should be deleted). The patch changes the order of
tasks. Calling DeleteMemberInDB before ListenerUpdate (i.e.
switch task 2 & 3) and thus the memebr is removed from HAProxy
List and info is updated to Amphorae as requested.
Change call to DeleteMemberInDB to take member and not member_id
change call to ListenerUpdate to use constants.LOADBALANCER, constants.LISTENER
Change-Id: I159831ca13fc8864798972f75d4b0c6e1fcecf26
Closes-Bug: #1514510
Once we add debug testenv, we can use "tox -e debug -- --debug"
to debug test cases when tox is running.
Change-Id: Ic0635f2b99c166c77757277e8273f42cdb17ce73
os.popen() is deprecated since version 2.6. Resolved with use of
subprocess module.
Closes-bug: #1529836
Change-Id: Ibdb8f13462e8416097006e19eb2214425364910e
When running the health manager on single host devstack machine, the
hosts IP cannot simply be used as if the health manager is running
on a host by itself. The solution here is to create a neutron port
on the lb-mgmt-net which sets up necessary ovs ports and also
reserves an IP. Then the script creates an ovs port for and sets
and interface with the information the neutron port create allocated.
There are some flaws with this but its a starting point and can and
probably will be improved on in the future. However it works for
now.
Change-Id: Ic3d3d1d63a5cc352c5fc00dea58bb16915754a7c
Closes-Bug: #1490033
We have an IdMixin. All the other objects with an ID here use it.
No point in inconsistently defining 'id' with the Amphora model.
Change-Id: I7758c08536e13d3ef653a604bcc74e1896c7d143
In _port_to_vip() which will be called by allocate_vip() it didn't
assign value to vip.load_balancer. This caused a exception in the
revert path in deallocate_vip(), which used vip.load_balancer.id
Change-Id: I31770e331edc1511385c3ba334afce6fc8af271c
Each config option has limitation for type and value.
In production code, oslo.conf can ensure user's input
is valid, but in unit test, test methods can pass if
we use method CONF.set_override without parameter
enforce_type=True even we pass wrong type or wrong
value to config option. This commit makes sure calling
method CONF.set_override with enforce_type=True.
Change-Id: I1e4f2aa5797ba8582a5637cd86c6b2b626daa923
Closes-Bug: #1517839
This makes more sense and also suppresses the error messages when
launching the service returned from oslo_messaing.get_rpc_server
service. Instead of that service wait() being called, the Consumer's
wait will be called.
Change-Id: I63816e92fbe26a4213946e6ab584531bdc3b7dd2
Closes-Bug: #1527418
There were some bad unit tests that would fail depending on the test
order. This would intermittently cause gate failures.
This patch corrects those unit tests.
Closes-Bug: #1526942
Change-Id: Id80a0a1a71a8248e5709f7df8c9af6a93a412f93
This patch fixes two things:
1. The sysctl settings were failing because some conntrack modules
were not loaded anymore.
2. I fixed the sysctl-set-value scripts to be able to handle
multi-value settings, such as tcp_rmem, in tripleo-image-elements[1].
Here I have removed the workaround we had in Octavia.
[1] https://review.openstack.org/#/c/134616/
Change-Id: Ib7ab4f487c1b792b70a110098bf7a28cb565ee55
Closes-Bug: #1527392
Current code to install diskimage-builder requirements in
devstack does not pass on enviroment settings to sudo, which can
cause the pip install to fail behind a proxy.
This happens when the user has http_proxy/https_proxy values set
but not for root. The stack.sh runs and installs for a while,
before failing on the Octavia diskimage-builder.
Change-Id: I196aca7004ea0eed713877e54ba10dcfe1d343e7
Closes-Bug: #1526108
