storage: update import path for Raft #29574
Conversation
|
This change is |
upstream moved from github.com/etcd to go.etcd.io/etcd. We're going to have to pick up an update soon to fix cockroachdb#28918, so it's easier to switch now. We're not picking up any new commits except for the renames. Release note: None
There was a problem hiding this comment.
Reviewable status:
complete! 0 of 0 LGTMs obtained (and 1 stale)
Gopkg.lock, line 1317 at r1 (raw file):
] pruneopts = "UT" revision = "1df1ddff4361ed7f2c0f33571923511889a115ce"
Should I worry that this is a different sha? Are we getting an etcd/raft bump at the same time?
There was a problem hiding this comment.
TFTR. There should be zero risk in this PR since we're not picking up any new code.
I'd also like to pick up the upstream fix when it's merged, though perhaps early in the 2.2 cycle.
In any case, the next time we bump Raft we'll have a good reason to do so, so it will be nice to not shave this yak then.
Reviewable status:
Gopkg.lock, line 1317 at r1 (raw file):
No, see
We're not picking up any new commits except for the renames.
We are picking up the renames, so the SHA is different. See https://github.com/etcd-io/etcd/commits/master/raft
There was a problem hiding this comment.
bors r=petermattis
Reviewable status:
|
👎 Rejected by code reviews |
There was a problem hiding this comment.
Dismissed @petermattis from a discussion.
Reviewable status:
There was a problem hiding this comment.
bors r=petermattis
Reviewable status:
|
👎 Rejected by code reviews |
|
bors r=petermattis |
29574: storage: update import path for Raft r=petermattis a=tschottdorf upstream moved from github.com/etcd to go.etcd.io/etcd. We're going to have to pick up an update soon to fix #28918, so it's easier to switch now. We're not picking up any new commits except for the renames. Release note: None 29585: gossip: avoid allocation of UnresolvedAddr in getNodeIDAddressLocked r=nvanbenschoten a=nvanbenschoten `getNodeIDAddressLocked` is called from `Dialer.ConnHealth` and `Dialer.DialInternalClient`. It was responsible for **1.71%** of all allocations (`alloc_objects`) on a 3-node long-running cluster that was running TPC-C 1K. Pointing into `nd.LocalityAddress` is safe because even if the `NodeDescriptor` itself is replaced in `Gossip`, the struct is never internally mutated. This is the same reason why taking the address of `nd.Address` was already safe. Release note (performance improvement): Avoid allocation when checking RPC connection health. Co-authored-by: Tobias Schottdorf <[email protected]> Co-authored-by: Nathan VanBenschoten <[email protected]>
Build succeeded |
| packages = ["."] | ||
| pruneopts = "UT" | ||
| revision = "bb4de0191aa41b5507caa14b0650cdbddcd9280b" | ||
|
|
There was a problem hiding this comment.
Um. This should not have gotten deleted as far as I can tell. Looking into it.
There was a problem hiding this comment.
Not your fault at all! That frustration was directed at dep.
There was a problem hiding this comment.
Well, I could've seen something unexpected was happening. :-)
upstream moved from github.com/etcd to go.etcd.io/etcd. We're going
to have to pick up an update soon to fix #28918, so it's easier to
switch now.
We're not picking up any new commits except for the renames.
Release note: None