-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ccl/sqlproxyccl: add --disable-connection-rebalancing flag to "mt start-proxy" #81712
ccl/sqlproxyccl: add --disable-connection-rebalancing flag to "mt start-proxy" #81712
Conversation
d9cd819
to
2d226ad
Compare
Strange. Couldn't seem to get the bazel tests to pass, but they all pass locally:
|
a7a0a34
to
7b91a21
Compare
Making this a draft to figure out what went wrong with the bazel tests. |
c0fc1f5
to
04eafbf
Compare
…rt-proxy" Previously, we added the connection rebalancing feature to the proxy, and that gets enabled automatically. This feature will not work with < v22.1 clusters since it relies on the session migration work that was added recently. In theory, the proxy will still work, but the balancer will constantly attempt to rebalance connection when we know that it will fail, and this adds to the overall latency of the connection since we suspend the processors during connection migration. To allow us to transition to v22.1 nicely, we will introduce a new flag `--disable-connection-rebalancing` to the `mt start-proxy` subcommand. When that flag is set, all connection rebalancing operations will be disabled. We have to roll out sqlproxy to CC before the clusters get their major upgrade since it contains some auth work that is a pre-req to v22.1. To avoid these latency issues, sqlproxy will be rolled out with the flag set. Once clusters have been upgraded to v22.1, the flag will be removed during startup. No release notes since `mt start-proxy` is internal only. Release note: None
04eafbf
to
0fcb822
Compare
It turned out that we had to bump the bazel test size from small to medium because we're hitting our timeout threshold. Hopefully tests are green this time. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Flaked on pkg/sql/logictest, which is unrelated to this PR. bors r=JeffSwenson |
TFTR! |
Build succeeded: |
Previously, we added the connection rebalancing feature to the proxy, and that
gets enabled automatically. This feature will not work with < v22.1 clusters
since it relies on the session migration work that was added recently. In
theory, the proxy will still work, but the balancer will constantly attempt
to rebalance connection when we know that it will fail, and this adds to the
overall latency of the connection since we suspend the processors during
connection migration.
To allow us to transition to v22.1 nicely, we will introduce a new flag
--disable-connection-rebalancing
to themt start-proxy
subcommand. Whenthat flag is set, all connection rebalancing operations will be disabled. We
have to roll out sqlproxy to CC before the clusters get their major upgrade
since it contains some auth work that is a pre-req to v22.1. To avoid these
latency issues, sqlproxy will be rolled out with the flag set. Once clusters
have been upgraded to v22.1, the flag will be removed during startup.
No release notes since
mt start-proxy
is internal only.Release note: None