From f86e1eddb5e1015ff39673b6616d5e745a570600 Mon Sep 17 00:00:00 2001 From: Kevin Carter Date: Tue, 4 Oct 2016 12:01:13 -0500 Subject: [PATCH] Update release note as the process has changed The bind mount for "/var/log" is only done on greenfield deployments, on upgrace this feature is disabled. Change-Id: I60bad26260f21316455b3e7a0c02a31069d08b1a Signed-off-by: Kevin Carter --- .../bindmount-logs-3c23aab5b5ed3440.yaml | 32 ++++++++----------- 1 file changed, 13 insertions(+), 19 deletions(-) diff --git a/releasenotes/notes/bindmount-logs-3c23aab5b5ed3440.yaml b/releasenotes/notes/bindmount-logs-3c23aab5b5ed3440.yaml index 1ceea65fa1..9a7cd487d4 100644 --- a/releasenotes/notes/bindmount-logs-3c23aab5b5ed3440.yaml +++ b/releasenotes/notes/bindmount-logs-3c23aab5b5ed3440.yaml @@ -1,25 +1,19 @@ --- features: - - Containers will now bind mount all logs to the physical host - machine in the "/openstack/log/{{ inventory_hostname }}" - location. This change will ensure containers using a block - backed file system (lvm, zfs, bfrfs) do not run into issues - with full file systems due to logging. + - In a greenfield deployment containers will now bind mount + all logs to the physical host machine in the + ``/openstack/log/{{ inventory_hostname }}`` location. This + change will ensure containers using a block backed file + system (lvm, zfs, bfrfs) do not run into issues with full + file disks due to logging. If this feature is not needed + or desired it can be disabled by setting the option + ``default_bind_mount_logs`` to *false*. upgrade: - - When upgrading deployers will need to ensure they have a - backup of all logging from within the container prior to - running the playbooks. If the logging node is present within - the deployment all logs should already be sync'd with the - logging server and no action is required. As a pre-step it's - recommended that deployers clean up logging directories from - within containers prior to running the playbooks. After the - playbooks have run the bind mount will be in effect at - "/var/log" which will mount over all previous log files and - directories. - - Due to a new bind mount at "/var/log" all containers will be - restarted. This is a required restart. It is recommended that - deployers run the container restarts in serial to not impact - production workloads. + - During an upgrade the option ``default_bind_mount_logs`` will + be set to *false*. This will ensure that an existing + deployment is not adversely impacted by container restarts. + If a deployer wishes to enable the default bind mount for + ``/var/log`` they can at a later date. fixes: - Logging within the container has been bind mounted to the hosts this reslves issue `1588051 _`