Helm Chart : Versions of dependencies unavailable

[B3nBeng]

Hi,

The versions specified (in Chart.yaml) of redis, memcached, rabbitmq and postgresql are not available anymore on Bitnami repo.

"Error: can't get a valid version for repositories memcached, rabbitmq, redis, postgresql. Try changing the version constraint in Chart.yaml"

[alexmv]

It looks like Bitnami dropped old releases from their index: bitnami/charts#10539.

We can either move to their GitHub-hosted archive:

diff --git kubernetes/chart/zulip/Chart.yaml kubernetes/chart/zulip/Chart.yaml
index b25b7bb..19f9d0f 100644
--- kubernetes/chart/zulip/Chart.yaml
+++ kubernetes/chart/zulip/Chart.yaml
@@ -15,22 +15,22 @@ version: 0.4.0
 appVersion: "6.1-0"
 dependencies:
   - name: memcached
-    repository: https://charts.bitnami.com/bitnami
+    repository: https://raw.githubusercontent.com/bitnami/charts/archive-full-index/bitnami
     tags:
       - memcached
     version: 6.0.16
   - name: rabbitmq
-    repository: https://charts.bitnami.com/bitnami
+    repository: https://raw.githubusercontent.com/bitnami/charts/archive-full-index/bitnami
     tags:
       - rabbitmq
     version: 8.32.0
   - name: redis
-    repository: https://charts.bitnami.com/bitnami
+    repository: https://raw.githubusercontent.com/bitnami/charts/archive-full-index/bitnami
     tags:
       - redis
     version: 16.8.7
   - name: postgresql
-    repository: https://charts.bitnami.com/bitnami
+    repository: https://raw.githubusercontent.com/bitnami/charts/archive-full-index/bitnami
     tags:
       - postgresql
     # Note: values.yaml overwrites posgresql image to zulip/zulip-postgresql:14

or, we can update the chart to the latest versions:

diff --git kubernetes/chart/zulip/Chart.yaml kubernetes/chart/zulip/Chart.yaml
index b25b7bb..79bad77 100644
--- kubernetes/chart/zulip/Chart.yaml
+++ kubernetes/chart/zulip/Chart.yaml
@@ -18,23 +18,23 @@ dependencies:
     repository: https://charts.bitnami.com/bitnami
     tags:
       - memcached
-    version: 6.0.16
+    version: 6.3.5
   - name: rabbitmq
     repository: https://charts.bitnami.com/bitnami
     tags:
       - rabbitmq
-    version: 8.32.0
+    version: 11.9.1
   - name: redis
     repository: https://charts.bitnami.com/bitnami
     tags:
       - redis
-    version: 16.8.7
+    version: 17.7.4
   - name: postgresql
     repository: https://charts.bitnami.com/bitnami
     tags:
       - postgresql
     # Note: values.yaml overwrites posgresql image to zulip/zulip-postgresql:14
-    version: 11.1.22
+    version: 11.9.13

 sources:
   - https://github.com/zulip/zulip

Can you try the latter and let us know if it works for you?

[B3nBeng]

Thanks you for the link of archives.
For the latest versions, it's works, i updated the Chart.yaml before opening this issue but i had few bugs with RabbitMQ so i wanted to try with the same versions specifies in the origin Chart.yaml but the version is not the problem.

[alexmv]

If the new versions work, we should bump our dependencies to them so we fix this for others.

[AdamJSoftware]

How did you end up fixing the rabbitmq issues?

[B3nBeng]

How did you end up fixing the rabbitmq issues?

Never fixed, we frequently encounter "client unexpectedly closed TCP connection" warn in the logs of the RabbitMQ pod.
This might not be an issue with Zulip but rather with the Kubernetes environment. The connections from the Zulip pod to the RabbitMQ pod appear to close and then reopen each time.

However, there is no impact; Zulip works perfectly for approximately 60 users each day.

2024-05-29 11:03:05.997881+00:00 [warn] <0.4118.722> closing AMQP connection <0.4118.722> (192.168.108.24:58176 -> 192.168.108.20:5672, vhost: '/', user: 'zulip'):
2024-05-29 11:03:05.997881+00:00 [warn] <0.4118.722> client unexpectedly closed TCP connection
2024-05-29 11:03:08.811423+00:00 [info] <0.6414.722> accepting AMQP connection <0.6414.722> (192.168.108.24:33608 -> 192.168.108.20:5672)
2024-05-29 11:03:08.813577+00:00 [info] <0.6414.722> connection <0.6414.722> (192.168.108.24:33608 -> 192.168.108.20:5672): user 'zulip' authenticated and granted access to vhost '/'

[timabbott]

I've definitely seen issues in the past with Docker/Kubernetes environments killing network connections that have been alive for more than 60 seconds or some other fixed period; that is likely the cause of the errors you're seeing in logs.

The Helm chart versions have been updated, so closing this issue as fixed.