This page lists common issues that may occur when running Milvus, as well as possible troubleshooting tips. Issues on this page fall into the following categories:
Boot errors are usually fatal. Run the following command to view error details:
$ docker logs <your milvus container id>
Errors that occur during runtime may cause service breakdown. To troubleshoot this issue, check compatibility between the server and your client before moving forward.
These issues occur during API method calls between the Milvus server and your client. They will be returned to the client synchronously or asynchronously.
The etcd cluster uses pvc by default. StorageClass needs to be preconfigured for the Kubernetes cluster.
When an etcd pod crashes with
Error: bad member ID arg (strconv.ParseUint: parsing "": invalid syntax), expecting ID in Hex, you can log into this pod and delete the
You can run the following code if multiple pods keeps crashing while
etcd-0is still running.
kubectl scale sts <etcd-sts> --replicas=1 # delete the pvc for etcd-1 and etcd-2 kubectl scale sts <etcd-sts> --replicas=3
When all pods crash, try copying the
https://github.com/etcd-io/bboltto modify database data.
All Milvus metadata are kept in the
keybucket. Back up the data in this bucket and run the following commands. Note that the prefix data in the
by-dev/meta/sessionfile does not require a backup.
kubectl kubectl scale sts <etcd-sts> --replicas=0 # delete the pvc for etcd-0, etcd-1, etcd-2 kubectl kubectl scale sts <etcd-sts> --replicas=1 # restore the backup data
If you need help solving a problem, feel free to: