minio

Highlights

• Support for compression + encryption. See (#11103) for more details.
• feat: migrate to ROOT_USER/PASSWORD from ACCESS/SECRET_KEY. See (#11185) for more details.
• Fix missing closure. See (#11237) for more details.
• Small file improvements. See (#11228, #11229) for more details.
• Fix: expire stale locks when owner is down. See (#11247) for more details.
• fix: treat errVolumeNotFound as EOF error in listPathRaw. See (#11238) for more details.
• Fix: initialize IAM store early to avoid any crash. See (#11236) for more details.
• Fixes in replication . See (#11227, #11243, #11244) for more details.
• fix: healing buckets during pool expansion. See (#11224) for more details.

Changelog

• Fix missing GetObjectNInfo Closure (#11243) (01/08/21) (Klaus Post)
• fix: print errors only when metacache status has errors (#11248) (01/08/21) (Harshavardhana)
• fix: readData in bulk call using messagepack byte wrappers (#11228) (01/07/21) (Harshavardhana)
• expire stale locks when owner is down (#11247) (01/07/21) (Harshavardhana)
• Pass versionID to checkReplicateDelete in web handler (#11244) (01/07/21) (Poorna Krishnamoorthy)
• fix: treat errVolumeNotFound as EOF error in listPathRaw (#11238) (01/07/21) (Harshavardhana)
• fix: getObject fd leaks in transition and replication code (#11237) (01/06/21) (Harshavardhana)
• initialize IAM store before Init() to avoid any crash (#11236) (01/06/21) (Harshavardhana)
• heal: Reduce whitespace ticker to 5 seconds (#11234) (01/06/21) (Anis Elleuch)
• fix: server/gateway banner formatting (#11230) (01/06/21) (Harshavardhana)
• update license verifier to use updated keys (#11197) (01/06/21) (Kanagaraj M)
• fix: use buffers only when necessary for io.Copy() (#11229) (01/06/21) (Harshavardhana)
• fix: rename remaining zone -> pool (#11231) (01/06/21) (Harshavardhana)
• Allow Compression + encryption (#11103) (01/05/21) (Klaus Post)
• Add Optimization as a type of change (#11221) (01/05/21) (Klaus Post)
• Pass deletemarker correctly to replicate opts (#11227) (01/05/21) (Poorna Krishnamoorthy)
• fix: healing buckets during pool expansion (#11224) (01/05/21) (Harshavardhana)
• tests: Fix occasional data race (#11223) (01/05/21) (Klaus Post)
• feat: migrate to ROOT_USER/PASSWORD from ACCESS/SECRET_KEY (#11185) (01/05/21) (Harshavardhana)

Highlights

• do not use large buffers if not necessary. See (#11220, #11213) for more details.
• Minor optimizations in handlers. See (#11211, #11217) for more details.
• remove/deprecate crawler disable environment. See (#11214) for more details.
• fix: calculate prometheus disks_offline/disks_total correctly. See (#11215) for more details.
• Optimizations to Get and Head object APIs. See (#11200, #11201, #11204, #11207) for more details.
• fix: admin trace throttled requests. See (#11192) for more details.

Changelog

• do not use large buffers if not necessary (#11220) (01/04/21) (Harshavardhana)
• handlers: Avoid initializing a struct in each handler call (#11217) (01/04/21) (Anis Elleuch)
• remove/deprecate crawler disable environment (#11214) (01/04/21) (Harshavardhana)
• fix: calculate prometheus disks_offline/disks_total correctly (#11215) (01/04/21) (Harshavardhana)
• tracing: NumSubscribers() to use atomic instead of mutex (#11219) (01/04/21) (Anis Elleuch)
• handlers: Little bit more optimizations (#11211) (01/04/21) (Anis Elleuch)
• erasure: avoid io.Copy in hotpaths to reduce allocation (#11213) (01/03/21) (Harshavardhana)
• feat: Small object optimization read data in single bulk call (#11207) (01/03/21) (Harshavardhana)
• parentDirIsObject() to return quickly with inexistant parent (#11204) (01/02/21) (Anis Elleuch)
• xl: Remove check-dir in ReadVersion (#11200) (01/02/21) (Anis Elleuch)
• fix: missing CopyObjectPart maxClients reorder (01/01/21) (Harshavardhana)
• trace: Reorder http/trace maxClients wrapping for correct tracing (#11202) (01/02/21) (Anis Elleuch)
• xl-storage: Do not stat bucket assuming the object exists (#11201) (01/01/21) (Anis Elleuch)
• fix: log once updating dataUsageCache versions (#11190) (12/31/20) (Harshavardhana)
• Fix caddy project url (#11198) (12/31/20) (Baptiste Mille-Mathias)
• fix: admin trace throttled requests (#11192) (12/30/20) (Ritesh H Shukla)
• Reduce logging when peer is offline (#11184) (12/30/20) (Ritesh H Shukla)
• update release Dockerfile string (12/30/20) (Harshavardhana)
• chore: enabled syntax highlighting in docs (#11182) (12/30/20) (0xflotus)
• fix: docs typo in object lock docs (#11181) (12/29/20) (Ravind Kumar)

Highlights

• Fix: Avoid crash in crawler when lifecycle is not initialized. See (#11170) for more details.
• Fixes and improvements in healing. See (#11171, #11173) for more details.
• Fix: re-arrange handlers to handle requests on /minio. See (#11177) for more details.

Changelog

• fix: enhance logging in crawler use console.Debug instead of logger.Info (#11179) (12/29/20) (Harshavardhana)
• fix: re-arrange handlers to handle requests on /minio (#11177) (12/28/20) (Harshavardhana)
• fix: heal optimizations in crawler to avoid multiple healing attempts (#11173) (12/28/20) (Harshavardhana)
• fix: use HealObject for cleaning up dangling objects (#11171) (12/27/20) (Harshavardhana)
• avoid a crash in crawler when lifecycle is not initialized (#11170) (12/26/20) (Harshavardhana)

Highlights

• Important performance improvements with list caching and memory consumption #11166, #11163, #11156

Changelog

• fix: inherit heal opts globally, including bitrot settings (#11166) (12/24/20) (Harshavardhana)
• fix: discarding results do not attempt in-memory metacache writer (#11163) (12/24/20) (Harshavardhana)
• webUI: change listing to 1000 keys from browser UI (#11159) (12/23/20) (Harshavardhana)
• Update lifecycle README docs (#11160) (12/23/20) (Poorna Krishnamoorthy)
• fix: docs typos and keywords (12/23/20) (Harshavardhana)
• fix: strict quorum in list should list on all drives (#11157) (12/23/20) (Harshavardhana)
• fix: re-use bytes.Buffer using sync.Pool (#11156) (12/22/20) (Harshavardhana)

Highlights

• List caching improvements. See (#11150) for more details.
• fix: select always online peers for remote listing. See (#11153) for more details.
• admin: ServerInfo returns info without object layer initialized. See (#11142) for more details.
• fix: change defaults for DNS cache dialer. See (#11145) for more details.
• fix: allow STS creds for admin accounts to add users. See (#11138) for more details.
• fix: delimiter based listing was broken without marker. See (#11136) for more details.
• replication: Ensure to always use nano precision source modtime. See (#11135) for more details.
• fix: reject overlapping object and prefix. See (#11130) for more details.

Changelog

• fix: allow HTTP2.0 to be always configured (12/22/20) (Harshavardhana)
• refactor cmd/crypto code for SSE handling and parsing (#11045) (12/22/20) (Andreas Auernhammer)
• fix: issues with handling delete markers in metacache (#11150) (12/22/20) (Harshavardhana)
• fix: select always online peers for remote listing (#11153) (12/22/20) (Harshavardhana)
• update x/net/http2 to address few bugs (#11144) (12/21/20) (Harshavardhana)
• Encrypt remote target if kms is configured (#11034) (12/21/20) (Poorna Krishnamoorthy)
• admin: ServerInfo returns info without object layer initialized (#11142) (12/21/20) (Anis Elleuch)
• fix: change defaults for DNS cache dialer (#11145) (12/21/20) (Harshavardhana)
• fix testMultipartUploadFailure to properly cleanup (#11137) (12/21/20) (Yingrong Zhao)
• normalize users with double // in accessKeys (#11143) (12/20/20) (Harshavardhana)
• fix: allow STS creds for admin accounts to add users (#11138) (12/19/20) (Harshavardhana)
• fix: support user/groups with '/' character (#11127) (12/19/20) (Harshavardhana)
• fix: delimiter based listing was broken without marker (#11136) (12/19/20) (Harshavardhana)
• fix: regression introduced in aws-sdk-go tests (12/18/20) (Harshavardhana)
• Profiling does not required object layer to be initialized (#11133) (12/18/20) (Anis Elleuch)
• replication: Ensure to always use nano precision source modtime (#11135) (12/18/20) (Anis Elleuch)
• fix: overlapping object and prefix rejected (#11130) (12/18/20) (Harshavardhana)

Highlights

• azure/s3 gateways: Pass ETag during GET call to avoid data corruption. See (#11024) for more details.
• fix: timer deadlock on expired timers. See (#11124) for more details.
• feat: treat /var/run/secrets/ on k8s as system cert directory. See (#11123) for more details.
• fix: reuser timers in erasure set hotpaths. See (#11106) for more details.

Changelog

• add _MINIO_SERVER_DEBUG env for enabling debug messages (#11128) (12/17/20) (Harshavardhana)
• fix: timer deadlock on expired timers (#11124) (12/17/20) (Harshavardhana)
• azure/s3 gateways: Pass ETag during GET call to avoid data corruption (#11024) (12/17/20) (Anis Elleuch)
• feat: treat /var/run/secrets/ on k8s as system cert directory (#11123) (12/16/20) (Harshavardhana)
• fix: reuser timers in erasure set hotpaths (#11106) (12/16/20) (Harshavardhana)
• fix testListMultipartUpload in aws-sdk-go (#11122) (12/16/20) (Yingrong Zhao)
• fix: avoid logging for Heal APIs in FS mode (#11121) (12/16/20) (Harshavardhana)

Highlights

• Healing fixes and improvements. See (#11098, #11097) for more details.
• fix: do not return nosuchversion error in deleteobjects API. See (#11114) for more details.
• Improvements to ILM. See (#11096) for more details
• fix: prioritize latest buckets to finish scanning faster. See (#11115) for more details.
• fix: remove inode free as part of SameDisk check. See (#11107) for more details.
• metacache improvements. See (#11102, #11109) for more details.

Changelog

• fix: prioritized latest buckets for crawler to finish the scans faster (#11115) (12/15/20) (Harshavardhana)
• fix: remove inode free as part of SameDisk check (#11107) (12/15/20) (Harshavardhana)
• metacache: Make very small requests transient (#11109) (12/15/20) (Klaus Post)
• fix: remove gorilla/rpc/ deps fully after our fork (#11108) (12/15/20) (Harshavardhana)
• Ignore ObjectNotFound errors in delete api while enforcing locking (#11114) (12/15/20) (Poorna Krishnamoorthy)
• Don't copy interesting ids, clean up logging (#11102) (12/14/20) (Klaus Post)
• fix: remove the requirement for healing buckets in ListBucketsHeal (#11098) (12/14/20) (Harshavardhana)
• lifecycle improvements and additional debug logging (#11096) (12/13/20) (Harshavardhana)
• fix: heal bucket metadata right before healing bucket (#11097) (12/13/20) (Harshavardhana)
• upgrade event notification dependencies (#11095) (12/12/20) (Harshavardhana)
• add vulnerability report policy (#11084) (12/13/20) (Andreas Auernhammer)
• xl: Always set root disk to true in test environment (#11094) (12/13/20) (Anis Elleuch)

Highlights

• No rolling upgrade supported in this release, please upgrade all servers at once.
• Important list caching improvements in #11087 #11078
• Major reduction in crawler memory usage #11068

Changelog

• return NoSuchVersion if invalid version-id is specified (#11091) (12/11/20) (Harshavardhana)
• metacache: Ask all disks when drive count is 4 (#11087) (12/12/20) (Anis Elleuch)
• fix: a crash when disk is nil, safe access on erasureDisks (#11089) (12/11/20) (Harshavardhana)
• Return version-id header in DeleteObject response (#11090) (12/11/20) (Poorna Krishnamoorthy)
• ignore NoSuchVersion error in DeleteObjects API (#11086) (12/11/20) (Poorna Krishnamoorthy)
• fix: TTFB prometheus metrics calculation (#11082) (12/10/20) (Harshavardhana)
• docs: fix the metrics formatting (#11081) (12/10/20) (Harshavardhana)
• update deps to latest for some vulnerable deps (#11080) (12/10/20) (Harshavardhana)
• crawler: Stream bucket usage cache data (#11068) (12/10/20) (Klaus Post)
• metacache: Speed up cleanup operation (#11078) (12/10/20) (Klaus Post)
• fix: refactor locks to apply them uniquely per node (#11052) (12/10/20) (Harshavardhana)
• fix: grafana double counting for bucket usage, histrogram and objects (#11070) (12/09/20) (Harshavardhana)
• ubi-minimal doesn't support arm32, remove from build manifest (12/09/20) (Harshavardhana)

Highlights

• Container: Move to Redhat UBI image. See (#11006) for more details.
• Improvements to locking. See (#11030) for more details.
• Add crawler delay config and dynamic config values. See (#11018, #11033) for more details.
• use argon2 with sync.Pool for better memory management. See (#11019) for more details.
• Add replication capacity metrics support in crawler. See (#10786) for more details.
• Add transition event notification. See (#11047) for more details.
• listcache: Add path index. See (#11063) for more details.
• reject mixed drive situations with drives on root disks. See (#11057) for more details.
• fix: grafana dashboard calculating online nodes. See (#11041) for more details.
• Add cache usage, total capacity to prometheus metrics. See (#11026) for more details.
• fix: heal multiple buckets in bulk. See (#11029) for more details.
• Stream bucket bandwidth measurements. See (#11014) for more details.

Changelog

• update nats.io/jwt dependency (#11066) (12/09/20) (Harshavardhana)
• listcache: Add path index (#11063) (12/09/20) (Klaus Post)
• federation: Redirect Lifecycle PUT request by bucket name (#11062) (12/09/20) (Anis Elleuch)
• reject mixed drive situations with drives on root disks (#11057) (12/09/20) (Harshavardhana)
• fix: grafana dashboard calculating online nodes (#11041) (12/09/20) (Nitish Tiwari)
• avoid updating nsswitch.conf for redhat UBI images (#11056) (12/08/20) (Harshavardhana)
• s3: Fix reading GET with partNumber specified (#11032) (12/08/20) (Anis Elleuch)
• Add documentation for bucket replication related metrics (#11055) (12/08/20) (Ritesh H Shukla)
• Remove alpine based image in favour or RedHat UBI (#11006) (12/09/20) (Nitish Tiwari)
• fix: auto update crawler meta version (12/08/20) (Harshavardhana)
• Revert "Add metrics for nodes online and offline (#11050)" (12/08/20) (Harshavardhana)
• make: Add hotfix target to generate hotfix binaries (#11053) (12/08/20) (Anis Elleuch)
• Add metrics for nodes online and offline (#11050) (12/08/20) (Ritesh H Shukla)
• add dynamic config docs (#11048) (12/07/20) (Harshavardhana)
• Add cache usage, total capacity to prometheus metrics (#11026) (12/07/20) (Poorna Krishnamoorthy)
• Add transition event notification (#11047) (12/07/20) (Poorna Krishnamoorthy)
• Add replication capacity metrics support in crawler (#10786) (12/07/20) (Ritesh H Shukla)
• fix: README with missing word (#11035) (12/07/20) (Justin Page)
• fix: re-use er.getDisks() properly in certain calls (#11043) (12/07/20) (Harshavardhana)
• fix: allow sub-admin to modify password for other users (#11039) (12/06/20) (Harshavardhana)
• fix: heal multiple buckets in bulk (#11029) (12/05/20) (Harshavardhana)
• support envs as well for new crawler sub-system (#11033) (12/04/20) (Harshavardhana)
• Add the dashboard json file (#11028) (12/05/20) (Nitish Tiwari)
• fix: release locks if the client timedout (#11030) (12/04/20) (Harshavardhana)
• Add crawler delay config + dynamic config values (#11018) (12/04/20) (Klaus Post)
• use argon2 with sync.Pool for better memory management (#11019) (12/03/20) (Harshavardhana)
• fix: update grafana dashboard docs (#11023) (12/04/20) (Nitish Tiwari)
• update gocredits with new updated dependencies (12/03/20) (Harshavardhana)
• fix: etcd import paths again depend on v3.4.14 release (#11020) (12/03/20) (Harshavardhana)
• Stream bucket bandwidth measurements (#11014) (12/03/20) (Ritesh H Shukla)

Highlights

• Do not use RELEASE.2020-12-03T00-03-10Z due to slowness seen during startup.
• Healing: Skip metacache entries during startup.
• Healing: Wait only if IOCount > 0 during startup.

Changelog

• skip metacache entries healing to speed up startup (12/02/20) (Harshavardhana)
• fix: startup being slow - wait only if IOCount > 0 (12/02/20) (Harshavardhana)
• fix: building docker images (12/02/20) (Harshavardhana)

Highlights

• Azure : load certs on windows from registry. See (#11016) for more details.
• Add support to make healing more aggressive. See (#11003) for more details.
• Adds "x-amz-usr-agent" and "x-id" headers in authentication of presignedURL. See (#10792) for more details.
• crawler: Missing folder heal check per set. See (#10876) for more details.
• select: Check if CSV is valid utf8. See (#10991) for more details.
• fix: allow admins to create users. See (#11005) for more details.
• Caching: Disallow writeback caching with cache_after. See (#11002) for more details.
• fix: avoid sending errors on missing objects on locked buckets. See (#10994) for more details.
• fix: allow accountInfo, addUser and getUserInfo implicit. See (#10978) for more details.
• fix: update RPM spec using rpmbuild. See (#10979) for more details.
• docker build related changes. See (#10983, #10958) for more details.

Changelog

• add support for tuning healing to make healing more aggressive (#11003) (12/02/20) (Harshavardhana)
• deprecate CommonName from TLS docs (#11017) (12/02/20) (Anis Elleuch)
• fix: load certs on windows from registry (#11016) (12/02/20) (Harshavardhana)
• Adds "x-amz-usr-agent" and "x-id" params to be used in authentication of presignedURL (#10792) (12/02/20) (ebozduman)
• update minio-go version v7.0.6 (#11012) (12/01/20) (Harshavardhana)
• Added set keyword in the command set to enable encryption on buckets (#11010) (12/01/20) (Sudarshan (Sid))
• rename server sets to server pools (12/01/20) (Harshavardhana)
• crawler: Missing folder heal check per set (#10876) (12/01/20) (Klaus Post)
• Revert "Support variable server sets (#10314)" (12/01/20) (Harshavardhana)
• select: Check if CSV is valid utf8 (#10991) (12/01/20) (Klaus Post)
• fix: allow admins to create users (#11005) (11/30/20) (Harshavardhana)
• Disallow writeback caching with cache_after (#11002) (11/30/20) (Poorna Krishnamoorthy)
• fix: s3select tests with new minio-py SDK (#10995) (11/29/20) (Harshavardhana)
• fix: avoid sending errors on missing objects on locked buckets (#10994) (11/28/20) (Harshavardhana)
• fix: allow accountInfo, addUser and getUserInfo implicit (#10978) (11/27/20) (Harshavardhana)
• fix: update RPM spec using rpmbuild (#10979) (11/27/20) (Harshavardhana)
• Upgrade compress and pgzip package (#10992) (11/27/20) (Klaus Post)
• add docker ubi buildx script (11/26/20) (Harshavardhana)
• fix: docker buildx support for multiplatform build (#10983) (11/26/20) (Harshavardhana)
• Add Dockerfile based on Red Hat UBI (#10958) (11/26/20) (Nitish Tiwari)
• Support variable server sets (#10314) (11/25/20) (Harshavardhana)

Highlights

• Add support for ILM transition and delete marker replication. See (#10928, #10892, #10565, #10878, #10396, #10945) for more details.
• Add support for replication of object tags, retention metadata. See (#10880) for more details.
• metacache improvements. See (#10973, #10965) for more details.
• fix: unwrapping in errors. See (#10949, #10934, #10917) for more details.
• Avoid sending errors in gob in storage requests. See (#10977) for more details.
• Add admin API to edit remote bucket target credentials. See (#10848) for more details.
• azure: Specify different Azure storage in the shell env. See (#10943) for more details.
• fix: do not return an error for successfully deleted dangling objects. See (#10938, #10961) for more details.
• fix: all hosts local and port same should be local erasure setup. See (#10951) for more details.
• add audit/admin trace support for browser requests. See (#10947) for more details.
• fix: do not crash PutObjectTags when node is down. See (#10940) for more details.

Changelog

• Avoid sending errors in gob in storage requests (#10977) (11/25/20) (Anis Elleuch)
• Refactor replication, ILM handling in DELETE API (#10945) (11/25/20) (Poorna Krishnamoorthy)
• metacache: Always close block writer (#10973) (11/25/20) (Klaus Post)
• fix: improve error handling in metacache (#10965) (11/25/20) (Harshavardhana)
• fix: marshaling stack overflow in noncurrentversion lifecycle config (#10971) (11/24/20) (Poorna Krishnamoorthy)
• Add admin API to edit remote bucket target credentials (#10848) (11/24/20) (Poorna Krishnamoorthy)
• run go mod tidy (11/24/20) (Harshavardhana)
• Remove msgp workaround (#10964) (11/24/20) (Klaus Post)
• Temporary workaround for msgp skipping (#10960) (11/23/20) (Klaus Post)
• fix: dangling object delete shall return object doesn't exist (#10961) (11/23/20) (Harshavardhana)
• azure: Specify different Azure storage in the shell env (#10943) (11/24/20) (Anis Elleuch)
• fix: do not return an error for successfully deleted dangling objects (#10938) (11/23/20) (Harshavardhana)
• fix: all hosts local and port same should be local erasure setup (#10951) (11/23/20) (Harshavardhana)
• fix: unwrapping issues with os.Is* functions (#10949) (11/23/20) (Harshavardhana)
• Show Delete replication status header (#10946) (11/21/20) (Poorna Krishnamoorthy)
• Remove platform specific structure definitions (#10935) (11/21/20) (Shireesh Anjal)
• Unwrap network errors (#10934) (11/20/20) (Klaus Post)
• add audit/admin trace support for browser requests (#10947) (11/20/20) (Harshavardhana)
• Rename OBD to Health (#10842) (11/21/20) (Shireesh Anjal)
• feat(docker): add a CI/CD Dockerfile, which starts the server right away (#10933) (11/20/20) (Dominik Lessel)
• docs: fix missing event types in notifications (#10944) (11/20/20) (Harshavardhana)
• fix: handle errors appropriately as they are wrapped (#10917) (11/20/20) (Harshavardhana)
• Display storage-class of transitioned object in HEAD (11/16/20) (Poorna Krishnamoorthy)
• fix: do not crash PutObjectTags when node is down (#10940) (11/20/20) (Harshavardhana)
• Add support for replication of object tags, retention metadata (#10880) (11/19/20) (Poorna Krishnamoorthy)
• validate service type of target in replication/ilm transition config (#10928) (11/19/20) (Poorna Krishnamoorthy)
• fix validation for deletemarker replication on object locked bucket (#10892) (11/12/20) (Poorna Krishnamoorthy)
• Add support for ILM transition (#10565) (11/12/20) (Poorna Krishnamoorthy)
• fix: delete marker replication should support directories (#10878) (11/12/20) (Harshavardhana)
• Revert "Revert "Add delete marker replication support (#10396)"" (11/19/20) (Harshavardhana)

Highlights

• listcache improvements. See (#10882, #10922, #10920) for more details.
• handle network shutdown properly. see (#10927) for more details.
• fix: check for gateway backend online without http request. See (#10924) for more details.
• fix: save healing tracker right before healing. See (#10915) for more details.
• WalkDir improvements. See (#10908, #10923, #10926) for more details.
• remove context deadline from Delete calls. See (#10901) for more details.
• fix: send replication completed event notification. See (#10902) for more details.
• fix: global in-place update boolean check. See (#10900) for more details.
• Fix disk cache usage percent for prometheus. See (#10898) for more details.
• listobjectversions: Add shortcut for Veeam blocks. See (#10893) for more details.

Changelog

• fix: network shutdown was not handle properly (#10927) (11/19/20) (Harshavardhana)
• fix: check for gateway backend online without http request (#10924) (11/19/20) (Harshavardhana)
• remove MinIO release tag as part of HTTP Server string (#10929) (11/19/20) (Harshavardhana)
• skip files that are not erasure objects or directories (#10926) (11/19/20) (Harshavardhana)
• allow prefix filtering for WalkDir to be optional (#10923) (11/18/20) (Harshavardhana)
• metacache: Allow prefix filtering (#10920) (11/18/20) (Klaus Post)
• Save listing error async (#10922) (11/18/20) (Klaus Post)
• fix: save healing tracker right before healing (#10915) (11/18/20) (Harshavardhana)
• increase readdir per block memory to facilitate faster WalkDir (#10908) (11/18/20) (Harshavardhana)
• update jstream to latest release v1.0.1 (#10909) (11/17/20) (Harshavardhana)
• listcache: Wrap returned errors (#10882) (11/17/20) (Klaus Post)
• remove context deadline from Delete calls (#10901) (11/17/20) (Harshavardhana)
• Use hw.physmem64 instead of hw.physmem for NetBSD in pkg/sys/getHwPhysmem (#10907) (11/17/20) (Matthias Petermann)
• fix: send replication completed event notification (#10902) (11/15/20) (Poorna Krishnamoorthy)
• fix: global in-place update boolean check (#10900) (11/15/20) (Rafael Bodill)
• Fix disk cache usage percent for prometheus (#10898) (11/14/20) (Poorna Krishnamoorthy)
• listobjectversions: Add shortcut for Veeam blocks (#10893) (11/13/20) (Klaus Post)
• fix: move context timeout closer to network for Delete calls (#10897) (11/13/20) (Harshavardhana)

Highlights

• Upgrade to Go 1.15.5 to address CVE-2020-28362, CVE-2020-28367 and CVE-2020-28366. Click here for more details.

Highlights

• Fix to sort directory objects correctly. See (#10886) for more details.
• fix: HDFS list bucket when subpath is provided. See (#10884) for more details.
• list cache improvements. See (#10879, #10883) for more details.
• fix: simplify healing metadata objects per set. See (#10867) for more details.
• DeletedObjects: Return objects on lock failure. See (#10874) for more details.

Changelog

• fix: disallow update if dates are same (#10890) (11/12/20) (Harshavardhana)
• Sort directory objects correctly (#10886) (11/12/20) (Klaus Post)
• docs: clarify notifications support for gateways (#10729) (11/12/20) (Omar Alvarez)
• Revert "Add delete marker replication support (#10396)" (11/12/20) (Harshavardhana)
• fix: HDFS list bucket when subpath is provided (#10884) (11/13/20) (cksac)
• ignore typed errors correctly in list cache layer (#10879) (11/12/20) (Harshavardhana)
• Allow deeper sleep (#10883) (11/12/20) (Klaus Post)
• Skip 0 byte stream writes (#10875) (11/11/20) (Klaus Post)
• fix: simplify healing metadata objects per set (#10867) (11/11/20) (Harshavardhana)
• DeletedObjects: Return objects on lock failure (#10874) (11/11/20) (Klaus Post)
• fix: bad example json for policy in replication docs (#10869) (11/11/20) (Anton Melser)
• Add delete marker replication support (#10396) (11/10/20) (Poorna Krishnamoorthy)

Highlights

• fix: negative STS IAM token TTL value. See (#10866) (11/10/20) for more details.
• fix: heal user/metadata right away upon server startup. See (#10863) for more details.
• re-route requests if IAM is not initialized. See (#10850) for more details.

Changelog

• fix: negative STS IAM token TTL value (#10866) (11/10/20) (Steven Reitsma)
• Reduce health check output (#10859) (11/10/20) (Klaus Post)
• fix: heal user/metadata right away upon server startup (#10863) (11/10/20) (Harshavardhana)
• fix: avoid divide by zero error single node distributed setup (#10862) (11/09/20) (Harshavardhana)
• doc: Add definition of 1KB and 1MB in prometheus (#10857) (11/09/20) (Anis Elleuch)
• Upgrade msgp to official release (#10858) (11/09/20) (Klaus Post)
• re-route requests if IAM is not initialized (#10850) (11/07/20) (Harshavardhana)

Highlights

• Rolling upgrades are not supported in this release.
• Add support for ListObject caching. See (#10814, #10648, #10784, #10844, #10836, #10837, #10834, #10820, #10809, #10800, #10802, #10826, #10825, #10824) for more details.
• fix: replaced drive properly by healing the entire drive. See (#10799) for more details.
• IAM related improvements. See (#10787) for more details.
• allow requests to be proxied when server is booting up. See (#10790) for more details.
• Optimize decryptObjectInfo. See (#10726) for more details.
• fix: add support for O_DIRECT reads for erasure backends. See (#10718) for more details.
• fix: honor delimiter as per AWS S3 spec. See (#10823) for more details.
• fix: use internode data for DisksInfo, VolsInfo in message pack. See (#10821) for more details.
• add missing TTL for STS credentials on etcd. See (#10828) for more details.
• fix: dangling objects on actual namespace. See (#10822, #10765) for more details.
• initialize IAM after etcd has initialized. See (#10819) for more details.
• rest client: Expect context timeouts for locks. See (#10782) for more details.
• optimize request URL encoding for internode. See (#10811) for more details.
• fix: infinite loop in cleanupStaleUploads of encrypted MPUs. See (#10845) for more details.
• fix: s3 gateway SSE pagination. See (#10840) for more details.
• Don't retain context in locker. See (#10515) for more details.
• Optimize FileInfo(Version) transfer. See (#10775) for more details.
• Add async caching commit option in diskcache. See (#10742) for more details.

Changelog

• fix: infinite loop in cleanupStaleUploads of encrypted MPUs (#10845) (11/06/20) (Steven Reitsma)
• metadata: Invalidate cache if unreadable and not updating (#10844) (11/06/20) (Klaus Post)
• fix: s3 gateway SSE pagination (#10840) (11/06/20) (Steven Reitsma)
• fix: add missing deleted key element in multiObjectDelete (#10839) (11/05/20) (Harshavardhana)
• metacache: Add option for life extension (#10837) (11/05/20) (Klaus Post)
• fix: dangling objects on actual namespace (#10822) (11/05/20) (Harshavardhana)
• Update msgp to main branch (#10835) (11/05/20) (Klaus Post)
• Don't replace caches that are receiving updates (#10834) (11/05/20) (Klaus Post)
• metacache: Always close the s2 writer (#10836) (11/05/20) (Klaus Post)
• add missing TTL for STS credentials on etcd (#10828) (11/04/20) (Harshavardhana)
• re-use optimized readdir for isDirEmpty() (#10829) (11/04/20) (Harshavardhana)
• fix: use internode data for DisksInfo, VolsInfo in message pack (#10821) (11/04/20) (Harshavardhana)
• Context based AccessKey passing (#10615) (11/04/20) (Bill Thorp)
• Add remote online/offline information (#10825) (11/04/20) (Klaus Post)
• Don't retain context in locker (#10515) (11/04/20) (Klaus Post)
• Keep transient lists while they are updating (#10826) (11/04/20) (Klaus Post)
• Add remote Diskinfo caching (#10824) (11/04/20) (Klaus Post)
• fix: honor delimiter as per AWS S3 spec (#10823) (11/04/20) (Harshavardhana)
• metacache: Add trashcan (#10820) (11/03/20) (Klaus Post)
• initialize IAM after etcd has initialized (#10819) (11/03/20) (Harshavardhana)
• use list cache for Walk() with webUI and quota (#10814) (11/03/20) (Harshavardhana)
• fix: IAM store fallback to list users and policies from disk (#10787) (11/02/20) (Harshavardhana)
• fix: move list quorum ENV to config (#10804) (11/02/20) (Harshavardhana)
• metacache: Check only one disk for updates (#10809) (11/02/20) (Klaus Post)
• Optimize FileInfo(Version) transfer (#10775) (11/02/20) (Klaus Post)
• Reduce WriteAll allocs (#10810) (11/02/20) (Klaus Post)
• optimize request URL encoding for internode (#10811) (11/02/20) (Harshavardhana)
• erasure: delete dangling objects automatically (#10765) (11/02/20) (Anis Elleuch)
• Add async caching commit option in diskcache (#10742) (11/02/20) (Poorna Krishnamoorthy)
• re-use remote transports in Peer,Storage,Locker clients (#10788) (11/02/20) (Harshavardhana)
• go mod update golang.org/x/text to address a CVE (#10805) (11/01/20) (Venkata K Sadineni)
• simplify monitoring doesn't need to be canceled (#10803) (10/31/20) (Harshavardhana)
• Recreate bucket metacache if corrupted (#10800) (10/31/20) (Klaus Post)
• Clean up metadata cache when deleting bucket (#10802) (10/31/20) (Klaus Post)
• fix: replaced drive properly by healing the entire drive (#10799) (10/31/20) (Harshavardhana)
• remove unnecessary logging and move to log once (#10798) (10/30/20) (Harshavardhana)
• allow requests to be proxied when server is booting up (#10790) (10/30/20) (Harshavardhana)
• fix: add support for O_DIRECT reads for erasure backends (#10718) (10/30/20) (Krishna Srinivas)
• Fix invalidated metacaches (#10784) (10/30/20) (Klaus Post)
• obd: Remove unused log constants (#10778) (10/29/20) (kannappanr)
• rest client: Expect context timeouts for locks (#10782) (10/29/20) (Klaus Post)
• Optimize decryptObjectInfo (#10726) (10/29/20) (Klaus Post)
• fix: handle a crash when AskDisks is set to -1 (#10777) (10/29/20) (Harshavardhana)
• fix: A possible crash when fi.Erasure.Distribution is empty (#10779) (10/28/20) (Harshavardhana)
• Add update retry limit and compare error by string instead (#10776) (10/28/20) (Klaus Post)
• fix: Do not cleanup range files in cache SaveMetadata when total hits are false (#10728) (10/28/20) (Kaloyan Raev)
• ListObjects Metadata Caching (#10648) (10/28/20) (Klaus Post)

Highlights

• Continuation of critical fixes made in the last release.
• Healing handles drive order mismatches. See (#10773) for more details.
• Fixes for CopyObject,PutObjectTagging APIs to ensure drive order, also make sure GetObject honors fallbacks due to drive order changes. See (#10772) for more details.

Changelog

• fix: heal object shards with ec.index and ec.distribution mismatches (#10773) (10/28/20) (Krishna Srinivas)
• fix: add more protection distribution to match EcIndex (#10772) (10/28/20) (Harshavardhana)
• Remove logging info from OBDInfoHandler (#10727) (10/28/20) (Shireesh Anjal)
• avoid caching http ranges in background goroutine (#10724) (10/27/20) (Kaloyan Raev)

Highlights

• Critical fixes related to replaced drives. See (#10748, #10759, #10764, #10766) for more details. Affects all releases since
  RELEASE.2020-09-05T07-14-49Z, if you are running any releases since RELEASE.2020-09-05T07-14-49Z it is recommended
  that you should schedule for an upgrade.
• Rolling upgrade is not supported (Internode API upgrade) (k8s/container deployments must upgrade all containers at once).
• Improvements to caching. See (#10762, #10763) for more details.
• Fix in Get/HeadObject to return 404 on dangling objects. See (#10753) for more details.
• IAM related improvements. See (#10719, #10700) for more details.
• Expire offline lockers. See (#10749) for more details.
• Aggressively slow down auto-healing. See (#10730) for more details.
• Object layer initialization sequence fix. See (10/19/20) for more details.
• s3 gateway DNS cache initialization fix. See (#10706) for more details.

Changelog

• fix: heal rejects objects with disk re-ordering issue (#10766) (10/26/20) (Krishna Srinivas)
• fix: use meta.Erasure.Index for GetObject() to reconstruct object (#10764) (10/26/20) (Krishna Srinivas)
• cache: rename function declarations (#10763) (10/26/20) (Poorna Krishnamoorthy)
• cache: fix call in GetObjectNInfo (#10762) (10/26/20) (Poorna Krishnamoorthy)
• fix: Get/HeadObject return 404 on non quorum objects (#10753) (10/26/20) (Anis Elleuch)
• fix: retain the previous UUID for newly replaced drives (#10759) (10/26/20) (Harshavardhana)
• fix: arm64 Dockerfile with missing \ for multiple ENVs (10/25/20) (Harshavardhana)
• turn-off checking for updates completely if MINIO_UPDATE=off (#10752) (10/24/20) (Harshavardhana)
• expire lockers if lockers are offline (#10749) (10/24/20) (Harshavardhana)
• make sure to preserve UUID from reference format (#10748) (10/24/20) (Harshavardhana)
• update STS examples to use latest v7 APIs (10/24/20) (Harshavardhana)
• erasure: Commit data before xl.meta in RenameData() (#10734) (10/24/20) (Anis Elleuch)
• tests: Avoid running TestNSRace in short test mode (#10735) (10/24/20) (Anis Elleuch)
• fix: slow down auto healing more aggressively (#10730) (10/22/20) (Harshavardhana)
• tests: Lower expectation in addr selection in rand cache dialer (#10739) (10/22/20) (Anis Elleuch)
• validate if iam store is initialized (#10719) (10/20/20) (Poorna Krishnamoorthy)
• docs: Update ILM doc with versioning features (#10714) (10/20/20) (Anis Elleuch)
• initialize object layer right after config has loaded (10/19/20) (Harshavardhana)
• initialize IAM as soon as object layer is initialized (#10700) (10/19/20) (Harshavardhana)
• fix: s3 gateway DNS cache initialization (#10706) (10/19/20) (Harshavardhana)

Highlights

• Improvements to bandwidth monitoring. See (#10683, #10698) for more details.
• Improvements to crawler. See (#10667, #10677, ) for more details.
• Healing related improvements. See (#10686, #10687, #10688) for more details.
• Add DNS cache support to avoid DNS flooding. See (#10693) for more details.
• update NTP package for accurate time resolution fixes. See (#10670) for more details.
• rename crawler config option to heal. See (#10678) for more details.
• Improvements to connecting to disks during startup. See (#10685, #10669) for more details.
• fix: Send delete marker creation event when appropriate. See (#10696) for more details.

Changelog

• fix: dex getting started guide URL (#10701) (10/18/20) (Márk Sági-Kazár)
• fix comment in bucket bandwidth package (10/17/20) (Harshavardhana)
• ilm: Send delete marker creation event when appropriate (#10696) (10/17/20) (Anis Elleuch)
• Clean up monitor on delete bucket (#10698) (10/16/20) (Ritesh H Shukla)
• add DNS cache support to avoid DNS flooding (#10693) (10/16/20) (Harshavardhana)
• Always close response body (#10697) (10/16/20) (Ritesh H Shukla)
• fix: azure gateway should reject bucket names with "." (#10635) (10/16/20) (ebozduman)
• fix: keep MRF channel open to avoid random CI crash (#10686) (10/16/20) (Klaus Post)
• fix: misc fixes for bandwidth reporting amd monitoring (#10683) (10/16/20) (Ritesh H Shukla)
• rename zones to serverSets to avoid terminology conflict (#10679) (10/15/20) (Harshavardhana)
• heal: Enable removing dangling delete markers (#10688) (10/15/20) (Anis Elleuch)
• fix: background heal rely on getOnlineDisks() (#10687) (10/15/20) (Harshavardhana)
• in getOnlineDisks wait for disks to be populated (#10685) (10/15/20) (Klaus Post)
• rename crawler config option to heal (#10678) (10/14/20) (Harshavardhana)
• fix: allow crawler to crawl on disks without usage constraints (#10677) (10/14/20) (Harshavardhana)
• update NTP package for accurate time resolution fixes (#10670) (10/13/20) (Harshavardhana)
• fix: move to go1.15 (#10660) (10/13/20) (Harshavardhana)
• fix: connect disks pre-emptively during startup (#10669) (10/13/20) (Harshavardhana)
• crawler: Remove waitForLowActiveIO (#10667) (10/13/20) (Klaus Post)
• remove deprecate readiness from healthcheck docs (#10659) (10/12/20) (Harshavardhana)
• fix: use the correct Action type for policy.Args and iampolicy.Args (#10650) (10/13/20) (飞雪无情)