Release Notes
InfluxDB Cluster v1.8.11-c1.2.0 版本
- chore: upgrade to go 1.21.13 (and support
GOMEMLIMITenvironment variable to set a soft memory limit includes the Go heap and all other memory managed by the runtime) - feat(gossip): gossip
/announcewill try to request other nodes when network failed - refactor(pool): optimize connection pool to reduce the number of
TCPandTIME_WAIT - refactor(meta-executor): optimize meta executor with connection pool
- feat(config): support
shard-reader-timeoutin data node configuration - chore(write): optimize logger output under write
timeoutfailed
- fix(cluster): fix nil pointer dereference in marshal binary of join cluster response #15
- fix(hinted-handoff): fix null pointer from hh caused panic #28
- fix(hinted-handoff): fix possible deadlock in node_processor #42
- fix(pool): fix possible deadlock in client_pool
- fix(copy-shard): fix read tcp i/o timeout in
copy-shard#39 - fix(meta-client): fix data race in meta client
- fix(show-shards): remove any expired shards and give empty expiry time for inf duration shards from the
/show-shardsoutput - fix(influxd): fix influxd close and reset issue
- fix(coordinator): fix closing opened twice in points writer
- fix(tsm1): Fix data race of seriesKeys in deleteSeriesRange
- chore(config): fix indent and typo in config.sample.toml and meta.config.sample.toml
InfluxDB OSS v1.8.11 版本
- #24764: chore: upgrade to influxdata/influxql v1.2.0
- #20793: feat: Add WITH KEY to show tag keys
- #21983: feat: SHOW TAG VALUES should produce results from one specific RP
- #22388: feat: show measurements database and retention policy wildcards
- #22620: feat: add thread-safe access to CountingWriter byte total
- #22638: feat: optionally dump queries to log on SIGTERM
- #23055: feat: configurable DELETE concurrency
- #23320: feat: log slow queries even without query logging
- #23425: feat: log the log level regardless of log level
- #23795: feat: add version number to debug/vars
-
#23978: feat: add the ability to log queries killed by
query-timeout - #18146: refactor(query): reuse matchAllRegex
- #19514: refactor: Use filepath.Walk
- #19323: refactor: Use binary.Read() instead of io.ReadFull()
- #22754: fix: detect misquoted tag values and return an error
- #17685: fix(tsm1): Fix temp directory search bug
- #18004: fix(httpd): Fixes key collisions when serializing /debug/vars
- #18129: fix(tsdb): Fix variables masked by a declaration
- #18429: fix(httpd): add option to authenticate prometheus remote read
- #19631: fix(CORS): allow PATCH
- #20694: fix: consistent error for missing shard
- #20802: fix(tsm1): fix data race and validation in cache ring
- #21558: fix: do not send non-UTF-8 characters to subscriptions
- #21592: fix: avoid rewriting fields.idx unnecessarily
- #21659: fix: do not close connection twice in DigestWithOptions
- #21666: fix: do not panic on cleaning up failed iterators
- #21792: fix: error instead of panic for statement rewrite failure
- #21795: fix: show shards gives empty expiry time for inf duration shards
- #22040: fix: copy names from mmapped memory before closing iterator
- #22038: test: fix order of index teardown
- #22273: fix: return correct count of ErrNotExecuted
- #22353: fix: correct error return shadowing
- #22348: fix: flux error properly read by cloud
- #22492: fix: For Windows, close temp file before removing
- #22566: fix(tsdb): sync series segment to disk after writing
- #22601: fix: create shards without overlaps
-
#22661: chore: fix deadlock in
influx_inspect dumptsi - #22703: fix: extend snapshot copy to filesystems that cannot link
- #22796: fix(restore): fix race condition which causes restore command to fail
- #21932: fix(tsi): sync index file before close
- #22929: fix: use defer for unlock
-
#22998: fix:
influxd-ctl backupwill create a working backup when only-shardgiven - #23217: fix: return underlying error creating a subscription
- #23197: fix(influx_inspect): multiple retention policies bug in influx_inspect export
- #21125: chore: fix typo in config.sample.toml
- #21746: fix(httpd): abort processing write request when encountering a precision error
- #23286: fix: MeasurementsCardinality should not be less than 0
- #23336: fix: remove data directory appending for influx_inspect verify
- #23387: fix: replace unprintable and invalid characters in errors
- #23396: fix: do not rename files on mmap failure
- #23430: fix: fully clean up partially opened TSI
- #23437: fix: remember shards that fail Open(), avoid repeated attempts
- #23461: fix: lost TSI reference / close TagValueSeriesIDIterator in error case
- #23467: fix: eliminate race condition on Monitor.globalTags
- #23532: fix: improve error messages opening index partitions
- #23539: fix: create TSI MANIFEST files atomically
- #23557: fix: add paths to tsi log and index file errors
- #23566: fix: add reporttsi to the help text
- #23580: fix: generalize test for Windows
- #23785: fix: use copy when a rename spans volumes
- #23787: fix: add tests for file rename across volumes
- #23822: fix: log errors in continuous query statistics storage
- #23727: fix: don't write skipped shard messages to the line protocol output destination
- #23916: fix: series file index compaction
- #24311: fix: do not escape CSV output
- #24509: fix: avoid SIGBUS when reading non-std series segment files
- #24565: fix: panic index out of range for invalid series keys
- #24784: fix: do not panic when empty tags are queried
- #24602: fix: improved shard deletion
- #24957: fix: ensure TSMBatchKeyIterator and FileStore close all TSMReaders
- #24975: fix: return MergeIterator.Close errors
- #25037: fix: ignore empty index error deleting last measurement
- #25082: fix: GROUP BY queries with offset that crosses a DST boundary fail.
- #25146: fix: Store.validateArgs wrongfully overwriting start, end unix time
- #25182: fix(tsi1): fix data race between appendEntry and FlushAndSync tsi1.(*LogFile)
- #22283: fix: require database authorization to see continuous queries
注意：为什么存在 v1.8.11？
因为 v1.8.10 是 1.8 的最后一个版本，已经长久不再维护。 v1.8.11 版本是在 v1.8.10 基础上，应用 1.9 至 1.11 版本中重要的 bug 修复补丁得到，commits 详见 https://github.com/chengshiwen/influxdb/compare/v1.8.10...1.8.11
- flux: 支持集群上的 flux query language
- influxd-ctl: 优化
influxd-ctl show-shards中的shard not found提示 - monitor: 优化统计信息中的
nodeID值为 tcp 地址
- influxql: 修复集群上的
show queries和kill query语句 - influxql: 修复集群上的
explain和explain analyze语句 - influxql: 修复集群上的
show measurement cardinality和show series cardinality语句
注: 所有 influxql (influx query language) 语句在集群上都已经支持
- 增加配置:
allow-out-of-order-writes和max-writes-pending - 升级 raft 库至 1.3.11
- meta handler: 修复 announcement map 并发写冲突的问题
- hinted-handoff: 优化启动时 hh segment 加载顺序
- hinted-handoff: 优化 hh segment 序列化，保持和 influxdb enterprise 格式一致
- hinted-handoff: 防止 partial write errors 缓存到 hinted handoff
- hinted-handoff: 当 field type errors 出现时，防止 hinted handoff 阻塞
- hinted-handoff: 修复 read bytes，blocked writes 和 dropped writes 未记录的问题
- hinted-handoff: 当 segment errors 出现时，确保 hh queue 正常往前运行
- hinted-handoff: 验证并截断 hh queue 中损坏的 segment 文件
- hinted-handoff: 优化 hinted handoff 统计指标
- hinted-handoff: 在 hh queue 增加 buffer 以避免 OOM
- influxd-ctl: 支持全局选项
-auth-type,-pwd,-secret和-user - influxd-ctl: 支持子命令
copy-shard,join,leave,remove-shard,show-shards,token和truncate-shards - influxd-ctl: remove-data 在 data 节点移除后，自动移除 data 目录
- influxd-ctl: remove-meta 在 meta 节点移除后，自动移除 meta 目录
- influxd-ctl: show 显示
version列用以健康检查 - config: 支持 meta 和 data 节点配置中共计 19 个设置项
- https: 支持 meta 和 data 节点的 https 通信
- auth: 支持 meta 和 data 节点的 jwt 和 basic 认证
- gossip: 支持 meta 和 data 节点的 gossip announcement
- query: 跳过持有查询所需分片的失败节点（如果另一个节点上有副本，它将在该节点上重试）
- hinted-handoff: 重构 hinted handoff 以确保清除在 node/shard/segment 目录结构下的数据
- hinted-handoff: 当数据节点被移除时，自动移除 hinted handoff 队列以避免手动清理
- chronograf: 适配 chronograf 访问 meta 服务的
/user和/role接口 - pool: 优化 shard writer 和 meta executor 中的连接池
- influx_inspect: 在 influx_inspect export 中导出多个保留策略的问题 #23197
- iterator: float, integer, string 和 boolean iterator finalized by GC
- query:
show measurements,show tag keys,show tag values,show field keys和show series只查询本地数据 - query:
/query接口中node_id未生效 - hinted-handoff: segment is full
- 发布 InfluxDB Cluster，其基于 InfluxDB Enterprise, InfluxDB v1.8.10 和 InfluxDB v0.11.1