-
Bug
-
Resolution: Done
-
High
-
Honolulu Release
-
None
-
None
When restarting postgres databases pods, cps data is lost.
Steps to reproduce:
1. Deploy cps helm chart and verify they are running properly
(⎈ |bruno.sakoto-ltec-com:rockhoppers)➜ kubernetes git:(review/claudio_david_gasparini/feature/CPS-175) ✗ kc get pods | grep cps bison-cps-56c7b5b6bf-f4j6n 1/1 Running 0 20m bison-cps-postgres-primary-7f44b99678-wzm6x 1/1 Running 0 20m bison-cps-postgres-replica-565b78c9f8-jwpxw 1/1 Running 0 20m
2. Create a dataspace and verify its presence in database
(⎈ |bruno.sakoto-ltec-com:rockhoppers)➜ kubernetes git:(review/claudio_david_gasparini/feature/CPS-175) ✗ kc exec -it bison-cps-postgres-primary-7f44b99678-wzm6x -- psql -d nsocpsdb -U nsocps -c "select * from dataspace;"
id | name
----+------------------
1 | bsa-dataspace-01
(1 row)
3. Delete postgres pods
(⎈ |bruno.sakoto-ltec-com:rockhoppers)➜ kubernetes git:(review/claudio_david_gasparini/feature/CPS-175) ✗ kc delete pods bison-cps-postgres-primary-7f44b99678-wzm6x bison-cps-postgres-replica-565b78c9f8-jwpxw pod "bison-cps-postgres-primary-7f44b99678-wzm6x" deleted pod "bison-cps-postgres-replica-565b78c9f8-jwpxw" deleted
4. Let postrgres pod restarting
(⎈ |bruno.sakoto-ltec-com:rockhoppers)➜ kubernetes git:(review/claudio_david_gasparini/feature/CPS-175) ✗ kc get pods | grep cps bison-cps-56c7b5b6bf-f4j6n 1/1 Running 1 27m bison-cps-postgres-primary-7f44b99678-btckl 1/1 Running 0 105s bison-cps-postgres-replica-565b78c9f8-plg76 1/1 Running 0 105s
Actual behaviour:
CPS data is lost
(⎈ |bruno.sakoto-ltec-com:rockhoppers)➜ kubernetes git:(review/claudio_david_gasparini/feature/CPS-175) ✗ kc exec -it bison-cps-postgres-primary-7f44b99678-btckl -- psql -d nsocpsdb -U nsocps -c "select * from dataspace;" id | name ----+------ (0 rows) (⎈ |bruno.sakoto-ltec-com:rockhoppers)➜ kubernetes git:(review/claudio_david_gasparini/feature/CPS-175) ✗ kc exec -it bison-cps-postgres-replica-565b78c9f8-plg76 -- psql -d nsocpsdb -U nsocps -c "select * from dataspace;" id | name ----+------ (0 rows)
Expected behaviour:
CPS data is not lost when postgres pods are restarted.
I'm recommending not fixing this bug in merge request 117099 currently in progress. Let's first have this large 117099 change merged and we'll fix this persistence bug in another change after.