)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":17068,"name":"Jean-Philippe Evrard","email":"openstack@a.spamming.party","username":"evrardjp"},"change_message_id":"e9ccc1447d760ff7792ff193fd1d66fe724b88ec","unresolved":false,"context_lines":[{"line_number":26,"context_line":"in case there is a pod restart, so it can recover itself and maintain all the data"},{"line_number":27,"context_line":"we had up to that point."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"Note that if the pod is rescheduled into another worker then we will hit the"},{"line_number":30,"context_line":"same issue, as the volume is local to the worker, we would get a chrashloopbackoff."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"task: 28729"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"bfb3d3c7_59a944dd","line":29,"range":{"start_line":29,"start_character":1,"end_line":29,"end_character":76},"updated":"2019-05-17 13:15:09.000000000","message":"In that case, shouldn\u0027t the cluster just resume operation by re-syncing with other workers? I am confused by what you mean here.","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":17068,"name":"Jean-Philippe Evrard","email":"openstack@a.spamming.party","username":"evrardjp"},"change_message_id":"bb1fbc23c07d07efd1f7aedb4599dcced8509890","unresolved":false,"context_lines":[{"line_number":26,"context_line":"in case there is a pod restart, so it can recover itself and maintain all the data"},{"line_number":27,"context_line":"we had up to that point."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"Note that if the pod is rescheduled into another worker then we will hit the"},{"line_number":30,"context_line":"same issue, as the volume is local to the worker, we would get a chrashloopbackoff."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"task: 28729"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"bfb3d3c7_7a2c3747","line":29,"range":{"start_line":29,"start_character":1,"end_line":29,"end_character":76},"in_reply_to":"bfb3d3c7_143871d1","updated":"2019-05-17 14:47:59.000000000","message":"Because these are stateful set, I supposed the pod would not be \"rescheduled\" into another worker, or that it wouldn\u0027t matter. You have clarified things for real cluster, as sync will happen, which is what I expected.","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":17068,"name":"Jean-Philippe Evrard","email":"openstack@a.spamming.party","username":"evrardjp"},"change_message_id":"2ee2da5ff77835a08a802d7b3acd7699fad3fa8f","unresolved":false,"context_lines":[{"line_number":26,"context_line":"in case there is a pod restart, so it can recover itself and maintain all the data"},{"line_number":27,"context_line":"we had up to that point."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"Note that if the pod is rescheduled into another worker then we will hit the"},{"line_number":30,"context_line":"same issue, as the volume is local to the worker, we would get a chrashloopbackoff."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"task: 28729"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"bfb3d3c7_d9bbf40b","line":29,"range":{"start_line":29,"start_character":1,"end_line":29,"end_character":76},"in_reply_to":"bfb3d3c7_5982a451","updated":"2019-05-17 13:20:29.000000000","message":"I got that part. This is changing the helm chart template, which means by default things run on /tmp for everyone, gates or not.\n\nI need to understand the impact on a real cluster before voting :)","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":23928,"name":"Pete Birley","email":"petebirley@gmail.com","username":"portdirect"},"change_message_id":"1563a396ede1bf7786eca1d56f1b9d801c92059d","unresolved":false,"context_lines":[{"line_number":26,"context_line":"in case there is a pod restart, so it can recover itself and maintain all the data"},{"line_number":27,"context_line":"we had up to that point."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"Note that if the pod is rescheduled into another worker then we will hit the"},{"line_number":30,"context_line":"same issue, as the volume is local to the worker, we would get a chrashloopbackoff."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"task: 28729"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"bfb3d3c7_5982a451","line":29,"range":{"start_line":29,"start_character":1,"end_line":29,"end_character":76},"in_reply_to":"bfb3d3c7_59a944dd","updated":"2019-05-17 13:15:51.000000000","message":"the state store is local in this case - so it wont migrate with the worker.","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":12281,"name":"Itxaka","email":"igarcia@suse.com","username":"itxaka"},"change_message_id":"0757c02e9f8e1986d85c9b03feece6a50bb2b03b","unresolved":false,"context_lines":[{"line_number":26,"context_line":"in case there is a pod restart, so it can recover itself and maintain all the data"},{"line_number":27,"context_line":"we had up to that point."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"Note that if the pod is rescheduled into another worker then we will hit the"},{"line_number":30,"context_line":"same issue, as the volume is local to the worker, we would get a chrashloopbackoff."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"task: 28729"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"bfb3d3c7_143871d1","line":29,"range":{"start_line":29,"start_character":1,"end_line":29,"end_character":76},"in_reply_to":"bfb3d3c7_d9bbf40b","updated":"2019-05-17 13:56:58.000000000","message":"we are talking about a 1 pod cluster here so not a real cluster. A real cluster (\u003e1 pod) will sync with other clusters properly and either has storage set up so the data is saved or it has storage disable in which case they dont care where the data goes as it would be reproduced.\n\nIMO having a cluster without storage is not something we want to support as it opens a can of worms (yo u lose everything if your whole cluster goes down because you dont have any real storage, only inside-container-storage)\n\nThis should only affect to 1 pod clusters with no storage enabled, in which case...you dont really care about where or how is your data saved. Nobody should use that, but we do on gates + some developers may be using it (me :P) so having your whole environment broken because mysql was restarted is pretty bad.","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":17068,"name":"Jean-Philippe Evrard","email":"openstack@a.spamming.party","username":"evrardjp"},"change_message_id":"209aa660bb9cad1011caade1806cc51771b9cc8e","unresolved":false,"context_lines":[{"line_number":9,"context_line":"There is currently an issue with deploying single pod mysql clusters"},{"line_number":10,"context_line":"in which restarting or killing the pod will result in a crashloopbackoff"},{"line_number":11,"context_line":"due to all the mysql data being lost and the start script thinking that"},{"line_number":12,"context_line":"the cluster was alive before, due to the grastate configmap, and trying to"},{"line_number":13,"context_line":"continue starting the cluster without bootstrapping it."},{"line_number":14,"context_line":""},{"line_number":15,"context_line":"Due to this, during our gating or testing if the mysql pod is killed or"},{"line_number":16,"context_line":"restarted, we will lose all the data, resulting in a broken environment."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"Even if we could fix the script to know that it should bootstrap the cluster"},{"line_number":19,"context_line":"again, our env will be broken as all the data from the rest of the openstack"},{"line_number":20,"context_line":"services is lost, so we would need to delete every chart and install it again"},{"line_number":21,"context_line":"so the databases are recreated and data populated. This is something that we"},{"line_number":22,"context_line":"cannot do during gating as its time consuming and we could end up in a loop never"},{"line_number":23,"context_line":"progressing further."},{"line_number":24,"context_line":""},{"line_number":25,"context_line":"Instead this patch will deploy a local volume for mysql so the data is kept intact"},{"line_number":26,"context_line":"in case there is a pod restart, so it can recover itself and maintain all the data"},{"line_number":27,"context_line":"we had up to that point."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"Note that if the pod is rescheduled into another worker then we will hit the"},{"line_number":30,"context_line":"same issue, as the volume is local to the worker, we would get a chrashloopbackoff."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"task: 28729"},{"line_number":33,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"bfb3d3c7_396fb0c6","line":30,"range":{"start_line":12,"start_character":0,"end_line":30,"end_character":83},"updated":"2019-05-17 13:12:05.000000000","message":"can you reduce the length of the lines here? It makes it harder to review :p","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":12281,"name":"Itxaka","email":"igarcia@suse.com","username":"itxaka"},"change_message_id":"0757c02e9f8e1986d85c9b03feece6a50bb2b03b","unresolved":false,"context_lines":[{"line_number":9,"context_line":"There is currently an issue with deploying single pod mysql clusters"},{"line_number":10,"context_line":"in which restarting or killing the pod will result in a crashloopbackoff"},{"line_number":11,"context_line":"due to all the mysql data being lost and the start script thinking that"},{"line_number":12,"context_line":"the cluster was alive before, due to the grastate configmap, and trying to"},{"line_number":13,"context_line":"continue starting the cluster without bootstrapping it."},{"line_number":14,"context_line":""},{"line_number":15,"context_line":"Due to this, during our gating or testing if the mysql pod is killed or"},{"line_number":16,"context_line":"restarted, we will lose all the data, resulting in a broken environment."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"Even if we could fix the script to know that it should bootstrap the cluster"},{"line_number":19,"context_line":"again, our env will be broken as all the data from the rest of the openstack"},{"line_number":20,"context_line":"services is lost, so we would need to delete every chart and install it again"},{"line_number":21,"context_line":"so the databases are recreated and data populated. This is something that we"},{"line_number":22,"context_line":"cannot do during gating as its time consuming and we could end up in a loop never"},{"line_number":23,"context_line":"progressing further."},{"line_number":24,"context_line":""},{"line_number":25,"context_line":"Instead this patch will deploy a local volume for mysql so the data is kept intact"},{"line_number":26,"context_line":"in case there is a pod restart, so it can recover itself and maintain all the data"},{"line_number":27,"context_line":"we had up to that point."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"Note that if the pod is rescheduled into another worker then we will hit the"},{"line_number":30,"context_line":"same issue, as the volume is local to the worker, we would get a chrashloopbackoff."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"task: 28729"},{"line_number":33,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"bfb3d3c7_d43179ee","line":30,"range":{"start_line":12,"start_character":0,"end_line":30,"end_character":83},"in_reply_to":"bfb3d3c7_396fb0c6","updated":"2019-05-17 13:56:58.000000000","message":"will do","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"}],"mariadb/templates/statefulset.yaml":[{"author":{"_account_id":17068,"name":"Jean-Philippe Evrard","email":"openstack@a.spamming.party","username":"evrardjp"},"change_message_id":"f99d8e484adab3745addf1b496de5c94df44afe3","unresolved":false,"context_lines":[{"line_number":239,"context_line":"            defaultMode: 0444"},{"line_number":240,"context_line":"        {{- if not .Values.volume.enabled }}"},{"line_number":241,"context_line":"        - name: mysql-data"},{"line_number":242,"context_line":"          hostPath: \"/tmp/mysql-data\""},{"line_number":243,"context_line":"        {{- end }}"},{"line_number":244,"context_line":"{{- if .Values.volume.enabled }}"},{"line_number":245,"context_line":"  volumeClaimTemplates:"}],"source_content_type":"text/x-yaml","patch_set":2,"id":"bfb3d3c7_795f68e5","line":242,"range":{"start_line":242,"start_character":21,"end_line":242,"end_character":36},"updated":"2019-05-17 13:22:09.000000000","message":"I am generally careful about putting data in /tmp. I would prefer if it would be, for the chart, into a configurable location, for which /tmp can be used in gates.\n\nBut I am waiting for comments on the commit message before voting.","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":23928,"name":"Pete Birley","email":"petebirley@gmail.com","username":"portdirect"},"change_message_id":"13d9d1c6d49a1fe368d9ee21d138d34ac65c5c89","unresolved":false,"context_lines":[{"line_number":239,"context_line":"            defaultMode: 0444"},{"line_number":240,"context_line":"        {{- if not .Values.volume.enabled }}"},{"line_number":241,"context_line":"        - name: mysql-data"},{"line_number":242,"context_line":"          hostPath: \"/tmp/mysql-data\""},{"line_number":243,"context_line":"        {{- end }}"},{"line_number":244,"context_line":"{{- if .Values.volume.enabled }}"},{"line_number":245,"context_line":"  volumeClaimTemplates:"}],"source_content_type":"text/x-yaml","patch_set":2,"id":"bfb3d3c7_f9491832","line":242,"range":{"start_line":242,"start_character":10,"end_line":242,"end_character":37},"updated":"2019-05-17 13:21:35.000000000","message":"i think we should probably drive this via a new key in the volume stanza of values, if set to null then we should use an emptydir as before.","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":12281,"name":"Itxaka","email":"igarcia@suse.com","username":"itxaka"},"change_message_id":"0757c02e9f8e1986d85c9b03feece6a50bb2b03b","unresolved":false,"context_lines":[{"line_number":239,"context_line":"            defaultMode: 0444"},{"line_number":240,"context_line":"        {{- if not .Values.volume.enabled }}"},{"line_number":241,"context_line":"        - name: mysql-data"},{"line_number":242,"context_line":"          hostPath: \"/tmp/mysql-data\""},{"line_number":243,"context_line":"        {{- end }}"},{"line_number":244,"context_line":"{{- if .Values.volume.enabled }}"},{"line_number":245,"context_line":"  volumeClaimTemplates:"}],"source_content_type":"text/x-yaml","patch_set":2,"id":"bfb3d3c7_b462c5d7","line":242,"range":{"start_line":242,"start_character":10,"end_line":242,"end_character":37},"in_reply_to":"bfb3d3c7_d969b48b","updated":"2019-05-17 13:56:58.000000000","message":"I struggled to get a key that expressed what this is....\n\nlocalHostPathSingleClusterPath ?\nSingleClusterLocalPath ?\nsingleClusterLocalPathMysqlData ?\n\nGoddammit...","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":17068,"name":"Jean-Philippe Evrard","email":"openstack@a.spamming.party","username":"evrardjp"},"change_message_id":"88061d7a9d5d7500c753811cb069e97fbce607c0","unresolved":false,"context_lines":[{"line_number":239,"context_line":"            defaultMode: 0444"},{"line_number":240,"context_line":"        {{- if not .Values.volume.enabled }}"},{"line_number":241,"context_line":"        - name: mysql-data"},{"line_number":242,"context_line":"          hostPath: \"/tmp/mysql-data\""},{"line_number":243,"context_line":"        {{- end }}"},{"line_number":244,"context_line":"{{- if .Values.volume.enabled }}"},{"line_number":245,"context_line":"  volumeClaimTemplates:"}],"source_content_type":"text/x-yaml","patch_set":2,"id":"bfb3d3c7_d969b48b","line":242,"range":{"start_line":242,"start_character":10,"end_line":242,"end_character":37},"in_reply_to":"bfb3d3c7_f9491832","updated":"2019-05-17 13:23:15.000000000","message":"haha, our reviews crossed. I think this is by far clearer than what I said in my review, but conveys EXACTLY what I wanted to say.\n\nThanks for clarifying","commit_id":"f544557ef7cf1f8cd3888c335094af9387d34acf"},{"author":{"_account_id":23928,"name":"Pete Birley","email":"petebirley@gmail.com","username":"portdirect"},"change_message_id":"7349143753eb4a00167a57b6a74dcb0844f660fb","unresolved":false,"context_lines":[{"line_number":240,"context_line":"        {{- if not .Values.volume.enabled }}"},{"line_number":241,"context_line":"        - name: mysql-data"},{"line_number":242,"context_line":"        {{- if .Values.volume.use_local_path_for_single_pod_cluster }}"},{"line_number":243,"context_line":"          hostPath: \"/tmp/mysql-data\""},{"line_number":244,"context_line":"        {{- else }}"},{"line_number":245,"context_line":"          emptyDir: {}"},{"line_number":246,"context_line":"        {{- end }}"}],"source_content_type":"text/x-yaml","patch_set":6,"id":"bfb3d3c7_56b8fc74","line":243,"range":{"start_line":243,"start_character":21,"end_line":243,"end_character":36},"updated":"2019-05-21 15:10:07.000000000","message":"it would be great if we could specify this - as /tmp will be wiped out on reboot.","commit_id":"06efdb19e3f011480b5885282be6c5984939c932"}]}