miner-genesis-init-container fail
julienpierini opened this issue · 9 comments
Hi,
I'm trying to deploy kuberneteth on my kubeadm cluster v1.10 but none miner containers can't run (it's in pending) because it wait for miner-genesis-init-container.
But my miner-genesis-init-container can't mkdir and touch in the volume mount because it haven't enough right : http://prntscr.com/j7sris
I don't know what i can do ?
@JPyml this might be a host specific issue, can you check if /var/lib/docker/geth-storage is writable? on the kubeadm host (not the container)
I just chmod and still the same issue : http://prntscr.com/j7t4zk
and i have deleted my pod
this is probably due to the usage of hostPath which is very limited, i wonder if it makes sense to switch to local at some point
I just chmod and still the same issue : http://prntscr.com/j7t4zk
can you try chown with whatever user is in charge of the container?
I run my kubernetes with my root account but I have try with my commun user but i have still the same issue
http://prntscr.com/j7tawn
@JPyml ok, i'll try to reproduce the issue. thanks for the report
It looks like this could be related to ethereum/go-ethereum#16464
The new docker image does not run with root permissions...
Dropping back to v1.8.3 of the ethereum/go-client image works 👍
It looks like this could be related to ethereum/go-ethereum#16464
The new docker image does not run with root permissions...
@Strydom good catch! thanks for the hint
@MaximilianMeister In node.toml.erb, Line 50 there's a typo
HTTPModules = ["db", "eth", "net", "web3", "personal", "web3"]
Replace the second web3 with "miner" and it worked for me.
EDIT: Never mind that, mining is still broken.
this is fixed in geth since release 1.8.6 see here
the breaking PR has been reverted
i've tested it on k8s 1.9.7 but I assume it is fixed as well on 1.10
@JPyml would be great if you can test it with a geth version >= 1.8.6 again, but i'll close the issue for now. feel free to reopen when it is still broken
thanks for the help