Mounter not using cluster DNS

[WallE256]

Hi,

I'm trying to use an in cluster object storage solution, and I'm unable to mount the volumes, even tho they are provisioned correctly. This is due to not using the cluster DNS. Can this be adjusted?

Example of error: 2023/06/05 09:52:41.159356 s3.ERROR code=RequestError msg=send request failed, err=Head "http://***.***.svc.cluster.local/cluster-volumes/pvc-7f661960-57c2-4e13-a9b5-9b393cf778f6/tlr9mntsgezlnlizpcjh4mkwka7wslev": dial tcp: lookup **.**.svc.cluster.local on 1.1.1.1:53: no such host

Another note, the Helm charts on cr.yandex are a version behind.

Thank you.

[vitalif]

Hi, csi-s3 doesn't override DNS settings, so it should probably use your cluster DNS without any change. Or does it require any special steps to use it? How is it set up?

[wethura]

Hi,

I'm trying to use an in cluster object storage solution, and I'm unable to mount the volumes, even tho they are provisioned correctly. This is due to not using the cluster DNS. Can this be adjusted?

Example of error: 2023/06/05 09:52:41.159356 s3.ERROR code=RequestError msg=send request failed, err=Head "http://***.***.svc.cluster.local/cluster-volumes/pvc-7f661960-57c2-4e13-a9b5-9b393cf778f6/tlr9mntsgezlnlizpcjh4mkwka7wslev": dial tcp: lookup **.**.svc.cluster.local on 1.1.1.1:53: no such host

Another note, the Helm charts on cr.yandex are a version behind.

Thank you.

remove the props: hostNetwork of daemonSet csi-s3, and the k8s will auto generate resolv.conf.

[vitalif]

Haha. Oops. It did override DNS then :D OK I removed it in master

[hwaastad]

Hi noticed this my self. On the host I see the following log:

geesefs[3656758]: caused by: Head "http://s3gw.s3gw.svc.cluster.local/csi-s3/pvc-bbad47f0-a538-46e1-868e-26d635fc854f/1nqjpswk47ixjp6ihjzoibcg2zvogxzc": dial tcp: lookup s3gw.s3gw.svc.cluster.local on <my-external-dns>:53: no such host

If I use the service ip, all is OK.

[helloqiu]

If geesefs runs by using systemd, it will use the DNS on the host rather than the DNS in cluster.