-
Notifications
You must be signed in to change notification settings - Fork 2.4k
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
Fix Redis Cluster issue during roll outs of new nodes with same addr #1914
Conversation
d344ec3
to
13f4f2d
Compare
This issue was surfaced in a Cloud Provider solution that used for rolling out new nodes using the same address (hostname) of the nodes that will be replaced in a Redis Cluster, while the former ones once depromoted as Slaves would continue in service during some mintues for redirecting traffic. The solution basically identifies when the connection could be stale since a MOVED response will be returned using the same address (hostname) that is being used by the connection. At that moment we consider the connection as no longer usable forcing to recycle the connection.
f1ce05a
to
9375185
Compare
Ill investigate why some tests are failing ... |
// of the connection. Force a DNS resolution when all connections | ||
// of the pool are recycled | ||
return true | ||
} |
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.
Replace else if
with else
and add return false
as a final clause. It should probably fix the tests...
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.
Moved to a switch, hope that this also fixes the broken tests.
Thanks 👍 |
Any chance to have a bugfix release version? |
Sure, I will push 8.11.4 |
* Upgrade redis-server version (#1833) * Upgrade redis-server version Signed-off-by: monkey <golang@88.com> * XAutoClaim changed the return value Signed-off-by: monkey <golang@88.com> * add cmd: geosearch, geosearchstore (#1836) * add cmd: geosearch, geosearchstore Signed-off-by: monkey92t <golang@88.com> * GeoSearchQuery and GeoSearchLocationQuery changed to pointer passing Signed-off-by: monkey92t <golang@88.com> * Added missing method XInfoStreamFull to Cmdable interface * Run go mod tidy in redisotel Signed-off-by: Bogdan Drutu <bogdandrutu@gmail.com> * Revert "ConnPool check fd for bad conns (#1824)" (#1849) This reverts commit 346bfaf. * Automate release process (#1852) * Bump github.com/onsi/gomega from 1.10.5 to 1.14.0 (#1832) * Bump github.com/onsi/gomega from 1.10.5 to 1.14.0 Bumps [github.com/onsi/gomega](https://github.com/onsi/gomega) from 1.10.5 to 1.14.0. - [Release notes](https://github.com/onsi/gomega/releases) - [Changelog](https://github.com/onsi/gomega/blob/master/CHANGELOG.md) - [Commits](onsi/gomega@v1.10.5...v1.14.0) --- updated-dependencies: - dependency-name: github.com/onsi/gomega dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com> * Upgrade gomega to v1.15.0 Signed-off-by: monkey92t <golang@88.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: monkey92t <golang@88.com> * Add version.go * Fix otel example * Fix package name in release script * More fixes for otel example * And more * Fix release.sh * Release v8.11.3 (release.sh) * Create an annotated tag to give release.yml chance to run * Tweak tag.sh * Add Cmd.Slice helper to cast to []interface{} (#1859) * after the connection pool is closed, no new connections should be added (#1863) * after the connection pool is closed, no new connections should be added Signed-off-by: monkey92t <golang@88.com> * remove runGoroutine Signed-off-by: monkey92t <golang@88.com> * pool.popIdle add p.closed check Signed-off-by: monkey92t <golang@88.com> * upgrade golangci-lint v1.42.0 Signed-off-by: monkey92t <golang@88.com> * Bump github.com/onsi/gomega from 1.15.0 to 1.16.0 (#1865) Bumps [github.com/onsi/gomega](https://github.com/onsi/gomega) from 1.15.0 to 1.16.0. - [Release notes](https://github.com/onsi/gomega/releases) - [Changelog](https://github.com/onsi/gomega/blob/master/CHANGELOG.md) - [Commits](onsi/gomega@v1.15.0...v1.16.0) --- updated-dependencies: - dependency-name: github.com/onsi/gomega dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * Add go 1.17 to the build matrix * Remove go 1.15 from build matrix * Add scan struct example (#1870) * Replace release job * Bump github.com/cespare/xxhash/v2 from 2.1.1 to 2.1.2 (#1872) Bumps [github.com/cespare/xxhash/v2](https://github.com/cespare/xxhash) from 2.1.1 to 2.1.2. - [Release notes](https://github.com/cespare/xxhash/releases) - [Commits](cespare/xxhash@v2.1.1...v2.1.2) --- updated-dependencies: - dependency-name: github.com/cespare/xxhash/v2 dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * Fix tag script to push tag by tag * Fix releasing.md * Fix/pubsub ping mutex (#1878) * Fix PubSub.Ping to hold the lock * Fix PubSub.Ping to hold the lock * add write cmd data-race test Signed-off-by: monkey92t <golang@88.com> Co-authored-by: monkey92t <golang@88.com> * chore: cleanup OpenTelemetry example * chore: gofmt all code * Refactor TestParseURL This is in preparation for supporting query parameters in ParseURL: - use an expected *Options instance to execute assertions on - extract assertions into helper function - enable parallel testing - condense test table * Add query parameter parsing to ParseURL() Before this change, ParseURL would only accept a very restricted set of URLs (it returned an error, if it encountered any parameter). This commit introduces the ability to process URLs like redis://localhost/1?dial_timeout=10s and similar. Go programs which were providing a configuration tunable (e.g. CLI flag, config entry or environment variable) to configure the Redis connection now don't need to perform this task themselves. * chore: add links to readme * chore: fix discussions link * empty hooks.withContext removed * chore: gofmt * chore: use conventional commits and auto-generate changelog * feat: add acl auth support for sentinels * chore: swap to acl auth at the test-level * Add support for BLMove command * chore: update dependencies * chore: update link * feat: add SetVal method for each command * feat: add Cmd.{String,Int,Float,Bool}Slice helpers and an example * chore: tweak GH actions to run all jobs * chore: add Lua scripting example * Fix Redis Cluster issue during roll outs of new nodes with same addr (#1914) * fix: recycle connections in some Redis Cluster scenarios This issue was surfaced in a Cloud Provider solution that used for rolling out new nodes using the same address (hostname) of the nodes that will be replaced in a Redis Cluster, while the former ones once depromoted as Slaves would continue in service during some mintues for redirecting traffic. The solution basically identifies when the connection could be stale since a MOVED response will be returned using the same address (hostname) that is being used by the connection. At that moment we consider the connection as no longer usable forcing to recycle the connection. * chore: lazy reload when moved or ask * chore: use conv commit message * chore: release v8.11.4 (release.sh) * fix: add whitespace for avoid unlikely colisions * fix: format * chore: fix links * chore: use ctx parameter in cmdInfo * Bump github.com/onsi/ginkgo from 1.16.4 to 1.16.5 (#1925) Bumps [github.com/onsi/ginkgo](https://github.com/onsi/ginkgo) from 1.16.4 to 1.16.5. - [Release notes](https://github.com/onsi/ginkgo/releases) - [Changelog](https://github.com/onsi/ginkgo/blob/master/CHANGELOG.md) - [Commits](onsi/ginkgo@v1.16.4...v1.16.5) --- updated-dependencies: - dependency-name: github.com/onsi/ginkgo dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * feat: add support for time.Duration write and scan * test: add test case for setting and scanning durations * chore: fix linter * fix(extra/redisotel): set span.kind attribute to client According to the opentelemetry specification this should always be set to client for database client libraries. I've also removed the SetAttributes call and instead set the attributes during creation of the span. This is what the library SHOULD be doing according to the opentelemetry api specification. * chore: update otel example * fix: update some argument counts in pre-allocs In some cases number of pre-allocated places in argument array is missing 1 or 2 elements, which results in re-allocation of twice as large array * chore: add example how to delete keys without a ttl * chore: don't enable all lints * chore(deps): bump github.com/onsi/gomega from 1.16.0 to 1.17.0 Bumps [github.com/onsi/gomega](https://github.com/onsi/gomega) from 1.16.0 to 1.17.0. - [Release notes](https://github.com/onsi/gomega/releases) - [Changelog](https://github.com/onsi/gomega/blob/master/CHANGELOG.md) - [Commits](onsi/gomega@v1.16.0...v1.17.0) --- updated-dependencies: - dependency-name: github.com/onsi/gomega dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com> * feat: Add redis v7's NX, XX, GT, LT expire variants * chore: add missing readme * chore: tweak feature links * chore: remove Discord * fix: set timeout for WAIT command. Fixes #1963 * build: update `go` directive in `go.mod` to 1.17 This commit enables support for module graph pruning and lazy module loading for projects that are at Go 1.17 or higher. Reference: https://go.dev/ref/mod#go-mod-file-go Reference: https://go.dev/ref/mod#graph-pruning Reference: https://go.dev/ref/mod#lazy-loading Signed-off-by: Eng Zer Jun <engzerjun@gmail.com> * chore: update link * chore: export cmder.SetFirstKeyPos to support build module commands * feat(redisotel): ability to override TracerProvider (#1998) * fix: add missing Expire methods to Cmdable This is a followup to #1928 * chore(deps): bump github.com/onsi/gomega from 1.17.0 to 1.18.1 Bumps [github.com/onsi/gomega](https://github.com/onsi/gomega) from 1.17.0 to 1.18.1. - [Release notes](https://github.com/onsi/gomega/releases) - [Changelog](https://github.com/onsi/gomega/blob/master/CHANGELOG.md) - [Commits](onsi/gomega@v1.17.0...v1.18.1) --- updated-dependencies: - dependency-name: github.com/onsi/gomega dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com> * Update README.md (#2011) chore: add fmt library in example code * chore: instrumentation name and version (#2012) * fix: invalid type assert in stringArg * chore: cleanup * fix: example/otel compile error (#2028) * fix: rename Golang to Go (#2030) https://go.dev/doc/faq#go_or_golang * feat: add support for passing extra attributes added to spans * feat: set net.peer.name and net.peer.port in otel example * chore: tweak Uptrace copy * feat: add support for COPY command (#2016) * feat: add support for acl sentinel auth in universal client * chore(deps): bump actions/checkout from 2 to 3 Bumps [actions/checkout](https://github.com/actions/checkout) from 2 to 3. - [Release notes](https://github.com/actions/checkout/releases) - [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md) - [Commits](actions/checkout@v2...v3) --- updated-dependencies: - dependency-name: actions/checkout dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com> * chore: add hll example * chore: tweak release script * chore: release v8.11.5 (release.sh) * chore: add discord back Co-authored-by: Eugene Ponizovsky <ponizovsky@gmail.com> Co-authored-by: Bogdan Drutu <bogdandrutu@gmail.com> Co-authored-by: Vladimir Mihailenco <vladimir.webdev@gmail.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Kishan B <kishancs46@gmail.com> Co-authored-by: Dominik Menke <dom@digineo.de> Co-authored-by: Gökhan Özeloğlu <gozeloglu@gmail.com> Co-authored-by: Justin Sievenpiper <justin@sievenpiper.co> Co-authored-by: Алексей Романовский <aromanovsky@epiphan.com> Co-authored-by: Stavros Panakakakis <stavrospanakakis@gmail.com> Co-authored-by: Pau Freixes <pfreixes@gmail.com> Co-authored-by: Ethan Hur <ethan0311@gmail.com> Co-authored-by: Jackie <18378976+Pyrodash@users.noreply.github.com> Co-authored-by: Kristinn Björgvin Árdal <kristinnardalsecondary@gmail.com> Co-authored-by: ffenix113 <razerer@bigmir.net> Co-authored-by: Bastien Penavayre <bastienPenava@gmail.com> Co-authored-by: James3 Li(李麒傑) <james3_li@asus.com> Co-authored-by: Eng Zer Jun <engzerjun@gmail.com> Co-authored-by: gzjiangtao2014 <gzjiangtao2014@corp.netease.com> Co-authored-by: Nelz <nelz9999@users.noreply.github.com> Co-authored-by: Daniel Richter <Nexyz9@gmail.com> Co-authored-by: Seyed Ali Ghaffari <ali.ghaffari@outlook.com> Co-authored-by: lintanghui <lintanghui@bilibili.com> Co-authored-by: hidu <duv123+github@gmail.com> Co-authored-by: Jonas Lergell <jonas.lergell@volvocars.com> Co-authored-by: Alex Kahn <alexanderkahn@gmail.com>
This issue was surfaced in a Cloud Provider solution that used for
rolling out new nodes using the same address (hostname) of the nodes
that will be replaced in a Redis Cluster, while the former ones once
depromoted as Slaves would continue in service during some mintues
for redirecting traffic.
The solution basically identifies when the connection could be stale
since a MOVED response will be returned using the same address (hostname)
that is being used by the connection. At that moment we consider the
connection as no longer usable forcing to recycle the connection.