pixie-io
Pixie version 0.1.7 can't be installed
Hi. I followed the instructions of Self-Hosted Pixie on Ubuntu and the LATEST_CLOUD_RELEASE was set to 0.1.7. I tried to deploy pixie but I fail on almost every step of the way:
- Deployment file (generated by kustomize) contains seccomp level of RuntimeProfile which causes the pthread_create() to fail (change it to Unconfined)
- Most of the images' path were broken (point to gcr.io/pixie-oss/pixie-dev/ which doesn't contain any images).
- There are no instructions of defining volumes.
These are the issues I manage to find and fix but still can't make pixie's pods to run.
**Steps to reproduce the behavior:
- Followed the instructions of Self-Hosted Pixie
- Make sure you deploy 0.1.7 version.
Expected behavior Clean installation and successful deployment.
App information (please complete the following information):
- Pixie version: 0.1.7
- K8s cluster version: client (1.27) and server (1.20)
- OS version: Ubunut 18 x86_64
Hi. I followed the instructions of Self-Hosted Pixie on Ubuntu and the LATEST_CLOUD_RELEASE was set to 0.1.7. I tried to deploy pixie but I fail on almost every step of the way:
- Deployment file (generated by kustomize) contains seccomp level of RuntimeProfile which causes the pthread_create() to fail (change it to Unconfined)
This is an issue with older docker runtimes. Your cluster is likely running a docker runtime before v20.10.12. I haven't noticed this in a cluster running docker runtime > v20.10.12. It should also work with containerd runtimes.
- Most of the images' path were broken (point to gcr.io/pixie-oss/pixie-dev/ which doesn't contain any images).
The image paths are overriden by kustomize, see https://github.com/pixie-io/pixie/blob/release/cloud/v0.1.7/k8s/cloud/public/kustomization.yaml If you follow the deploy instructions at https://docs.px.dev/installing-pixie/install-guides/self-hosted-pixie/ the correct images should be output by kustomize.
- There are no instructions of defining volumes.
Unfortunately volume deployment depends on your k8s cluster provider. I'd suggest looking at documentation for where/how the k8s cluster is created.
These are the issues I manage to find and fix but still can't make pixie's pods to run.
Are there any other issues here? With a lack of concrete logs or issues, I can't offer any more help.
Thanks @vihangm for your answer.
- This is my docker version: Docker version 20.10.12, build e91ed57
- When I run this command: kustomize build k8s/cloud_deps/public | kubectl apply -f - I get the following error: error: resource mapping not found for name: "pl-nats-pdb" namespace: "plc" from "STDIN": no matches for kind "PodDisruptionBudget" in version "policy/v1" ensure CRDs are installed first (when I change policy/v1 to policy/v1beta1 it passes with no error).
- After running the command: kustomize build k8s/cloud/public/ | kubectl apply -f -, this is the pods' status: NAME READY STATUS RESTARTS AGE api-server-68fbdbbf48-rtzrv 0/1 CrashLoopBackOff 4 3m20s artifact-tracker-server-67c7f96866-782n6 1/1 Running 0 3m20s auth-server-7566ccc7cc-6nslq 0/1 CrashLoopBackOff 4 3m20s cloud-proxy-856dc76bc7-wdk2j 0/2 CrashLoopBackOff 10 3m20s config-manager-server-64fc49c769-d6v6v 1/1 Running 0 3m20s cron-script-server-6b89df5599-dz7pc 0/1 CrashLoopBackOff 4 3m20s hydra-5865f8cb9d-zh8g5 0/2 Init:CrashLoopBackOff 4 3m20s indexer-server-6dfbb9f5fb-8s6s6 0/1 CrashLoopBackOff 5 3m20s kratos-d8dbd7956-wf444 0/2 Init:Error 5 3m20s metrics-server-6b5b8ffc59-k69h7 0/1 CrashLoopBackOff 5 3m20s pl-elastic-es-data-0 0/1 Pending 0 11m pl-elastic-es-master-0 0/1 Pending 0 11m pl-elastic-es-master-1 0/1 Pending 0 11m pl-nats-0 0/1 Pending 0 11m pl-nats-1 0/1 Pending 0 11m pl-nats-2 0/1 Pending 0 11m plugin-db-updater-job-7wdj6 0/1 Init:0/1 0 3m20s plugin-server-65d79d4d5-2b96g 0/1 CrashLoopBackOff 4 3m19s postgres-7875d7f995-4tp92 0/1 Pending 0 11m profile-server-748bcddd68-9fm8n 0/1 Running 5 3m19s project-manager-server-85c5cccfcf-9hfvj 0/1 CrashLoopBackOff 4 3m19s scriptmgr-server-65cf5ffb88-clm9j 0/1 CrashLoopBackOff 4 3m19s vzconn-server-5fd45bdc58-brcwm 0/1 CrashLoopBackOff 5 3m19s vzmgr-server-6f64485649-682lj 0/1 CrashLoopBackOff 4 3m18s
- Attaching the kustomization.yaml file.
Thanks. kustomization.zip
- Looks like you are using a k8s cluster older than v1.21 since PDB has been available in
policy/v1since v1.21. I'd recommend updating since v1.21 was EOL a year ago, it's likely other things won't work on a cluster that old either. - The pixie pods are likely crashing due to nats being unavailable. You'd need to debug that first. And if the pods still crash the pods logs would be necessary to debug.