update GCP client libraries #249
Conversation
skriss
force-pushed
the
update-gcp-lib
branch
2 times, most recently
from
2292e3b to
6c0fa2b
Compare
|
We can go ahead with review/merge for this PR. I've tested everything on a GKE cluster -- @nrb would be nice to also run through e2e if possible. |
|
I can double check. If you're looking for PV testing, though, that likely isn't ready. Last I checked, the Kubicorn GCP profile didn't have the service account integration necessary to manage cloud provider objects. |
|
|
||
| return ret, nil | ||
| if obj.Prefix != "" { | ||
| res = append(res, obj.Prefix[0:strings.LastIndex(obj.Prefix, delimiter)]) |
There was a problem hiding this comment.
I'm not sure we still need to take the substring? According to https://godoc.org/cloud.google.com/go/storage#Query:
// ... Objects whose names,
// aside from the prefix, contain delimiter will have their name,
// truncated after the delimiter, returned in prefixes.
There was a problem hiding this comment.
the Prefix values always contain a trailing delimiter (just re-confirmed this), so we need to remove it to get backup names. The language in that godoc comment is super-confusing, but I'm fairly certain that how this works is:
- if there are any objects in the bucket that don't contain
delimiterin their name, they'll be returned, with theNamefield set and noPrefix. (note that we don't have any of these in Ark because all of our objects are prefixed with<backup-name>/). - for any objects that do contain
delimiter, a result will be returned with theNamefield empty andPrefixset to the substring of their name up to and includingdelimiter. - results containing only
Prefixare de-duped.
There was a problem hiding this comment.
Ah, truncated after the delimiter. OK.
Signed-off-by: Steve Kriss <steve@heptio.com>
|
Can you split out the dep/vendor changes into a separate commit? |
Signed-off-by: Steve Kriss <steve@heptio.com>
Signed-off-by: Steve Kriss <steve@heptio.com>
|
split. |
oadp-1.0: Makefile.prow: implement verify, travis pin go 1.16
…vmware-tanzu#249) * Update velero library and deployment version to 1.14 Also bump go version and builder to 1.22.3, kubevirtci version and kubevirt provider to 1.29 Signed-off-by: Shelly Kagan <skagan@redhat.com> * Run make gomod-update Signed-off-by: Shelly Kagan <skagan@redhat.com> * Use crictl instead of docker to pull from quay kubevirt/kubevirtci#1100 Signed-off-by: Shelly Kagan <skagan@redhat.com> * Fix cluster-sync due to velero 1.14 changes * As stated in velero 1.14 release docs: `CSI plugin has been merged into velero repo in v1.14 release. It will be installed by default as an internal plugin, and should not be installed via "–plugins " parameter in "velero install" command.` As such removed the installation of velero-plugin-for-csi * Velero added a confirmation check when adding a plugin when the velero pod is already running, hence needed to add the `confirm` flag. * Increased pod memory request and limit due to OOMKilled with smaller amount. * updated aws plugin to v1.10 version Signed-off-by: Shelly Kagan <skagan@redhat.com> * Update ginkgo cli in tests to 2.17.1 Signed-off-by: Shelly Kagan <skagan@redhat.com> --------- Signed-off-by: Shelly Kagan <skagan@redhat.com>
No description provided.