Hi!
And here is the final release of 1.25.0 and just copied the feature
description over from rc.1,
but wanted to note an additional important change/bugfix with error-report
listing.
We now a limit parameter for error-reports list queries and can now report
the total count
of error-reports without having to fetch their header data. This should
vastly improve
the prometheus /metrics queries and also get rid of 10s timeout errors in
the logs.
The biggest feature in this release is that Linstor got aware of disk
failures detected by DRBD.
If a disk fails now and DRBD switches the resource to a diskless, Linstor
will mark those
resource with a skip-disk flag and will add the --skip-disk argument to
drbd commands.
Without this feature Linstor would mostly fail on most commands because it
couldn't adjust
drbd resources and even unrelated resources might have become problematic.
We also added IP_TOS socket options to improve Linstor network
communication on congested networks.
## [1.25.0] - 2023-10-25
### Changed
- Added ErrorReport when sending a ping request fails
- Sos-report's tar command no longer use --verbose
- ErrorReports list now works with paging and a default limit.
Fixing node timeouts on large error-reports
- Resource-group adjust warns now for placement count needs to be reduced
(instead of error)
### Fixed
- Fixed setting 'on-no-quorum' to 'suspend-io' at the resource-group level
does not propagate down to existing resources
- Fixed logic in setLogLevel to prevent error log if no error happened
- Fixed possible 'Access to deleted Resource' in the AutoDiskfulTask
- Fixed possible left over during AutoDiskfulTask's cleanup of excess
resources
## From rc1
### Added
- Added automated addition of --skip-disk to drbdadm adjust when a disk
failure is detected
- Added first version of "effective_props" (currently only used for
"DrbdOptions/SkipDisk", other props will follow)
- Added attempt in setting IP_TOS to LOW_DELAY (works in java11 or with
forced IPv4)
- Added creation time to ApiCallRc
### Changed
- Improved checks for "OtherController" response
- Do not send fetchFreeSpaces to diskless-only nodes
- Improved responses during tiebreaker takeover
### Fixed
- Fixed possible FileWatch leak
- Fixed toggle-disk now properly setting StorPoolName
- Delete node-connection when last property gets deleted
- Properly delete DRBD_DELETE flag during undelete ('r c' on a DELETING
resource)
- Fixed regex-pattern for parsing "dmsetup ls [--target TARGET]" command
- Fixed typo in /v1/action[s]/snapshot/multi
- Fixed incorrectly documented return type of "snapshot create-multi"
- Fixed NPE in schedules etcd driver
https://pkg.linbit.com//downloads/linstor/linstor-server-1.25.0.tar.gz
Linstor PPA:
https://launchpad.net/~linbit/+archive/ubuntu/linbit-drbd9-stack
Cheers,
Rene
And here is the final release of 1.25.0 and just copied the feature
description over from rc.1,
but wanted to note an additional important change/bugfix with error-report
listing.
We now a limit parameter for error-reports list queries and can now report
the total count
of error-reports without having to fetch their header data. This should
vastly improve
the prometheus /metrics queries and also get rid of 10s timeout errors in
the logs.
The biggest feature in this release is that Linstor got aware of disk
failures detected by DRBD.
If a disk fails now and DRBD switches the resource to a diskless, Linstor
will mark those
resource with a skip-disk flag and will add the --skip-disk argument to
drbd commands.
Without this feature Linstor would mostly fail on most commands because it
couldn't adjust
drbd resources and even unrelated resources might have become problematic.
We also added IP_TOS socket options to improve Linstor network
communication on congested networks.
## [1.25.0] - 2023-10-25
### Changed
- Added ErrorReport when sending a ping request fails
- Sos-report's tar command no longer use --verbose
- ErrorReports list now works with paging and a default limit.
Fixing node timeouts on large error-reports
- Resource-group adjust warns now for placement count needs to be reduced
(instead of error)
### Fixed
- Fixed setting 'on-no-quorum' to 'suspend-io' at the resource-group level
does not propagate down to existing resources
- Fixed logic in setLogLevel to prevent error log if no error happened
- Fixed possible 'Access to deleted Resource' in the AutoDiskfulTask
- Fixed possible left over during AutoDiskfulTask's cleanup of excess
resources
## From rc1
### Added
- Added automated addition of --skip-disk to drbdadm adjust when a disk
failure is detected
- Added first version of "effective_props" (currently only used for
"DrbdOptions/SkipDisk", other props will follow)
- Added attempt in setting IP_TOS to LOW_DELAY (works in java11 or with
forced IPv4)
- Added creation time to ApiCallRc
### Changed
- Improved checks for "OtherController" response
- Do not send fetchFreeSpaces to diskless-only nodes
- Improved responses during tiebreaker takeover
### Fixed
- Fixed possible FileWatch leak
- Fixed toggle-disk now properly setting StorPoolName
- Delete node-connection when last property gets deleted
- Properly delete DRBD_DELETE flag during undelete ('r c' on a DELETING
resource)
- Fixed regex-pattern for parsing "dmsetup ls [--target TARGET]" command
- Fixed typo in /v1/action[s]/snapshot/multi
- Fixed incorrectly documented return type of "snapshot create-multi"
- Fixed NPE in schedules etcd driver
https://pkg.linbit.com//downloads/linstor/linstor-server-1.25.0.tar.gz
Linstor PPA:
https://launchpad.net/~linbit/+archive/ubuntu/linbit-drbd9-stack
Cheers,
Rene