2020-02-17 10:36:32 UTC - gfouquier: Hi everyone, I am trying a new attemp to move pulsar logs out of syslog (v2.5.0), but I still got the same issue: my configuration is taken into account (I check with the log4j debug mode) and log files are created where i want, but remain desperately empty. "pulsar.log.dir" had been configured in log4j.yaml file and the RollingFile appender is the default appender. I also configured the log conf and dir in pulsar_env, pulsar_tools_env and bkenv. Does anyone known what I missed ? ---- 2020-02-17 13:37:49 UTC - Santiago Del Campo: The url provided is the one provided by default ---> <http://broker:8080> ---- 2020-02-17 13:39:24 UTC - Santiago Del Campo: that name (broker), inside any pod of the cluster resolves to thw correct namespace (For example... if broker it's deployed inside *pulsar1* namespace, any other service within that same namespace will resolve "broker" to the pod of the broker inside *pulsar1* namespace). ---- 2020-02-17 13:45:43 UTC - Santiago Del Campo: This is the output of the last two commands you mentioned: ---- 2020-02-17 13:46:40 UTC - Santiago Del Campo: If i do a nslookup to the broker name.. i also get this: ---- 2020-02-17 13:47:08 UTC - Santiago Del Campo: There seems to be some sort of mismatch, right? :thinking_face: ---- 2020-02-17 13:49:34 UTC - Santiago Del Campo: And finally.. this is the env variable passed when the pulsar-dashboard it's deployed: ---- 2020-02-17 17:18:51 UTC - Sijie Guo: how do you start pulsar components? how do you configure the settings? ---- 2020-02-17 17:21:47 UTC - Sijie Guo: `bin/pulsar-admin clusters get local` it seems that you somehow have `broker.default.svc.cluster.local` stored in the cluster configuration for cluster `local`. I would suggest you using `bin/pulsar-admin clusters update` to update the cluster configuration. ---- 2020-02-17 17:51:55 UTC - Santiago Del Campo: Mm.. i understand, i'll do that and get back to you with the results. ---- 2020-02-17 18:03:06 UTC - Santiago Del Campo: Great... that did the trick, much cleaner this way... although i still have to redeploy the pulsar-dashboard, i suppose i did something wrong during deployment or somehow in the configuration store the default namespace it's written by default :thinking_face: ---- 2020-02-17 18:48:44 UTC - Ming: @David Kjerrumgaard Thanks David. I noticed two of your PRs opened years back. I am glad/hope #6335 will be merged soon in 2.6. +1 : David Kjerrumgaard ---- 2020-02-17 19:08:07 UTC - Sijie Guo: How do you run pulsar metadata setup? ---- 2020-02-17 19:12:25 UTC - Santiago Del Campo: I grab the cluster-metadata.yaml and point it to the *pulsar1* namespace (kubectl create -f cluster-metadata.yaml --namespace=*pulsar1*). ---- 2020-02-17 19:14:20 UTC - Santiago Del Campo: Ooooh right... the cluster metadata sets the fqdns for the broker.... wow, seems never paid enough attention to it.... ---- 2020-02-17 19:14:51 UTC - Santiago Del Campo: Well, i got it now... i appreciate your help and your patience :pray: ---- 2020-02-18 03:11:48 UTC - Kelvin Sajere: So Pulsar works on websocket, binary protocol and http . • The rest http is mainly for admin e.g localhost:8080/_admin/v2/clusters_ • binary is for communication with broker e.g <pulsar://localhost:6650> • Websocket is for librarys not yet supported e.g <ws://192.168.99.109:8080/ws/v2/producer/persistent/public/default/my-topic> The error you get means you are hitting the pulsar service but theres no reply corresponding to that URL. You can try it out with websocket on pulsar 2.4.2 using WebSocket King client: Postman for WebSockets chrome plugin . You can also read more about this under client library section <https://pulsar.apache.org/docs/en/client-libraries-java/> ---- 2020-02-18 03:21:04 UTC - Kelvin Sajere: If you need to use a web UI, this is a seperate project called apache pulsar manager . <https://github.com/apache/pulsar-manager> ---- 2020-02-18 03:42:57 UTC - Kelvin Sajere: ```docker run -it -p 9527:9527 -e REDIRECT_HOST=<http://192.168.99.109> -e REDIRECT_PORT=9527 -e DRIVER_CLASS_NAME=org.postgresql.Driver -e URL='jdbc:<postgresql://127.0.0.1:5432/pulsar_manager>' -e USERNAME=pulsar -e PASSWORD=pulsar -e LOG_LEVEL=DEBUG --mount source=data,target=/data apachepulsar/pulsar-manager:v0.1.0 /bin/sh ``` This is the way i run it on my local system from powershell. The `---mount` will create the folder if it doesn't exit. i don't think `-v` will. I mine on docker tool. Either way `REDIRECT_HOST` == localhost ---- 2020-02-18 06:59:12 UTC - gfouquier: I start broker and bookkeeper using the pulsar binary, and the pulsar command is called from systemd. ---- 2020-02-18 06:59:50 UTC - gfouquier: the following properties had been defined in log4j2.yaml ---- 2020-02-18 07:00:32 UTC - gfouquier: the log dir and log conf is also configured in pulsar-env.sh, pulsar-tools.sh and bkenv.sh ---- 2020-02-18 07:06:20 UTC - gfouquier: and here is log4j debug logs ---- 2020-02-18 07:06:28 UTC - gfouquier: thanks ! ----
