[doc] Remove guidance to drain RMQ which can result in failures
oslo_messaging isn't set up to handle this situation and appears to expect the node to come back (as would happen during a version upgrade). As a result, client processes don't attempt to re-create queues which lived on the node being upgraded. This situation doesn't auto-recover which can be very disruptive. Change-Id: I608202d665417682ce7309bb9bf4b52a2cf4373e
This commit is contained in:
parent
32e503af74
commit
4abddeaf30
@ -117,18 +117,6 @@ Warnings
|
||||
Deploying Infrastructure Hosts
|
||||
==============================
|
||||
|
||||
#. Drain RabbitMQ connections (optional)
|
||||
|
||||
In order to cleanly hand over connections from one member of the RabbitMQ
|
||||
cluster to another, the instance being reinstalled should be drained.
|
||||
This can be achieved by running the following from the instance to be
|
||||
reinstalled and waiting for the RabbitMQ admin interface to indicate that
|
||||
socket descriptors have reduced to zero.
|
||||
|
||||
.. code:: console
|
||||
|
||||
rabbitmq-upgrade drain
|
||||
|
||||
#. Disable HAProxy back ends (optional)
|
||||
|
||||
If you wish to minimise error states in HAProxy, services on hosts which are
|
||||
|
Loading…
Reference in New Issue
Block a user