- now works with default configuration
- updated readme
- fixed unknown field "spec.terminationGracePeriodSeconds" in deployment.yaml
error. This field should be defined in pod spec not in deployment spec
- improved how aliasgroups are defined in helm chart
- change default to kubernetes "Deployment" instead of "StatefulSet"
- added support for kubernetes hostAliases
Signed-off-by: Rash419 <rashesh.padia@collabora.com>
Change-Id: Ief1e2b3f0ea130425853babbd28462d48e50280f
Instead of using URL parameter hash to direct all the users on same document to same server,
use stick-tables for it
original problem:
Problem was load balancer provided us with required persistence and balance both.
we directed traffic based on the url_param which means,
all the traffic with same url_param will go to same pod/server.
All the new values of url_param will be directed to different servers in round-robin way
this can be less efficient in a high traffic environment
solution:
this new method allows us to balance and manage persistence separately
so we can specify our desired algorithm for load balancing
and persistence is managed using stick-table
Signed-off-by: Pranam Lashkari <lpranam@collabora.com>
Change-Id: I80d63a2fc33f796c4534d94fb8de1e44c62ee9e2
updated the instructions to setup online in kubernetes
Signed-off-by: Pranam Lashkari <lpranam@collabora.com>
Change-Id: Ic01d94d095eac9dba54a09351eabcbc6ba83c16b
* switch monitoring path to REST endpoint for admin metrics
* update service type to ClusterIP from NodePort
* adding environment variables for deployment customization
* switch /tmp folder mapping from memory to disk
Change-Id: I78dab256343c20bf84060bb4fbae203d4769be35
Reviewed-on: https://gerrit.libreoffice.org/85217
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
Tested-by: Michael Meeks <michael.meeks@collabora.com>
