Leofs Versions Save

The LeoFS Storage System

1.4.4

3 years ago

1.4.3

5 years ago

CHANGELOG

New Features

  • #420 S3-API leo_gateway Support object compression during network transfers

Fixed Bugs

  • #1082 leo_storage Fail to start in case it receives delete-(bucket|dir) messages during its startup
  • #1089 leo_gateway No error log when setting an invalid IP of one of LeoManager nodes
  • #1091 leo_storage Rebalance get slow down when ingesting messages into leo_rebalance_queue
  • #1093 leo_storage No error log when setting an invalid IP of one of LeoManager nodes
  • #1116 s3-api leo_gateway Cyberduck not listen files - Fix LeoFS' S3-API
  • #1133 s3-api leo_gateway PUT Object tagging MUST be ignored

Improvements

  • #965 leo_manager create|import-user can fail with records partially updated
  • #1079 f_ring Dump RING before changing it
  • #1086 leo_storage Setting a large num_of_containers causes leo_storage not to work properly
  • #1090 leo_manager whereis output can be wrong when the storage node is slow - Revise the format of whereis command's output
  • #1094 leo_gateway Reading a file through s3fs can fail
  • #1103 leo_gateway rest-api GET with the Range header doesn't work
  • #1123 snmp leo_storage SNMP don't return leo_per_object_queue
  • #1130 leofs Make it work with OTP 21
  • #1152 leo_manager Prohibit any operations which change the cluster status while the RING inconsistency happens

Libraries

Leo Project

Others

1.4.3-rc1

5 years ago

1.4.2

5 years ago

CHANGELOG

Fixed Bugs

  • #947 [leo_object_storage] MD5 check doesn't happen in case GET with Range header with is_strict_check = true
  • #1030 [leo_object_storage] Fix error handling/where to call of file:datasync in case of SYNC_MODE_WRITETHROUGH
  • #1046 [rack aware] Doesn't work as expected
  • #1047 [rack aware] rack-id isn't displayed on leofs-adm status <NODE>
  • #1049 [leo_storage][multipart] Abort multi-part upload leaves parts objects
  • #1050 [leo_gateway][multipart] Return wrong status code in some abnormal cases
  • #1058 [leo_mq] the state is not changed to suspend after issuing mq-suspend
  • #1059 [leo_mq] The state suspend can be changed to running/idle by leo_watchdog

Improvements

  • #549 [leo_object_storage][compaction] Timeout for safe
  • #618 [leo_manager][leo_storage] How to recover from a queue being lost
  • #766 [leo_gateway] Revise the default for large_object.reading_chunked_obj_len
  • #824 [libcutil] Suppress debug info to STDOUT
  • #1016 [leo_manager][leo_storage] leofs-adm recover-file doesn't work for a large object
  • #1038 [leofs-adm] leofs-adm issue when requesting msgs to LeoFS on Debian v9.x
  • #1060 [leo_mq] Improve the CPU utilization by changing consumers to event-driven way
  • #1061 [leo_mq] enables users to issue mq-suspend in case the state is idle
  • #1068 [leo_manager] log_level is not output on json I/F
  • #1069 [leo_manager] The response from get-(users|endpoints|buckets) are inconsistent when it's empty

Libraries

Leo Project

Others

1.4.1

6 years ago

CHANGELOG

Fixed Bugs

LeoGateway

  • #1019 Wrong mime type set for webm & webp
  • #1021 [S3-API] PUT Object acl should be ignored
  • #1031 [REST-API] CDN integration hasn't worked

LeoManager

  • #1003 [multi-dc replication] To avoid executing recover-cluster in case of not using multi DC replication

Improvements

  • #673 [gateway] Graceful configuration reload
  • #891 [manager] Administrative port on managers shouldn't listen on all interfaces
  • #995 [eleveldb] Make log files less fragmented
  • #1018 [libcutil] Fix subunit dependencies for the upcoming Ubuntu-18.04
  • #1035 [leofs-adm] To be able to correctly terminate leofs-adm's commads on Ubuntu-18.04

Libraries

Leo Project

Others

1.4.0

6 years ago

CHANGELOG

Fixed Bugs

LeoGateway

  • #894 Immature Body can cause some problem

LeoManager

  • #892 delete-bucket can stop proceeding for a long time
  • #949 Suppress error messages in crash.log during mnesia initialization
  • #964 import-user with the access-key-id belonging to a deleted user doesn't work
  • #989 Should return 503 in case leo_watchdog get triggered
  • #1003 To avoid executing recover-cluster in case of not using multi DC replication
  • #1010 The whereis command doesn't work for an object having grandchildren

LeoStorage

  • #543 read-repair don't work when metadata is valid but avs is broken
  • #546 recover-file don't work when metadata is valid but avs is broken
  • #553 GET can be false negative more than necessary
  • #722 Errors on gateway/storage when uploading lots of files
  • #758 leo_backend_db leo_object_storage du/compact-status can get stuck long time while fetching objects
  • #764 delete-bucket take much time due to unnecessary congestion
  • #960 Suppress error logs on stop
  • #963 Suppress error logs on start
  • #966 Node entered "strange" state with internal crashes
  • #975 leo_objct_storage Avoid gen_server crash due to timeout
  • #1011 The delete API (S3-API) doesn't delete grandchildren
  • #1014 data-compaction After overwriting a large file and executing data-compaction, cannot retrieve the one
  • #1017 data-compaction Grandparent (root) objects can be removed unintentionally

Others

  • #971 leo_redundant_manager Keep dumping the same error in case mq worker process died
  • #977 Handle supervisor restart properly (All Leo's components)

Improvements

  • #779 To be able to build LeoFS with Erlang/OTP 20
  • #840 Support "notify" type of services for systemd
  • #941 leo_storage Making more information available over SNMP for storage nodes
  • #955 Supervisor received unexpected message appeared in error log on startup
  • #976 Refactor slow processing notification to avoid bottlenecks
  • #980 Tune the default 30 min for TimeoutStartSec
  • #983 leofs-adm Add recover-disk to improve the recover performance in case of disk drive failures
  • #994 leo_backend_db Implement iterater_move in batch

Libraries

Leo Project

Others

1.4.0-rc2

6 years ago

LeoFS v1.4.0's pre-release