Skip to content

Releases: ceph/ceph-nvmeof

1.4.1

01 Dec 19:00
2cbe91b
Compare
Choose a tag to compare

yield subsystem_namespace_count #957
Separate namespaces limit to a total and per subsystem one. #956
spdk: support the option of no huge pages #898
Enforce scale limitations #952
Give an explicit error message in case of creating too many namespaces on a subsystem Fixes bz#2282295
Fix the limit on the number of namespaces with namespace masking #950
catch concrete exceptions #947
Move inband authentication controller key from host to subsystem. #937
Allow any address for listeners using address 0.0.0.0 (or :: for IPv6). #935
Allow using inband authentication with TLS. #933
control/grpc.py: osdwait logic #931
Add a CLI command to change host DHCHAP keys. #924
create listeners after subsystem #927
Move the Prometheus startup to the end of the server initialization. #919
Update SPDK to version 24.09. Fixes #917
Abort gateway in case of SPDK ping failure. Fixes #914
main exit (do not catch System exit, eturn exit code 0 from TERM signal handler) #912
Allow saving SPDK and moniotr client outputs to a file. Fixes #906
Fix problems using IPv6 addresses in connect and discover. Fixes #903
Add a separate log level field for SPDK, don't display error for string context params. Fixes #873 Fixes #901
Remove "RBD Pool" column fron "namespace list" output. Fixes #899.
Limit number of not-visible namespaces to 1000. Fixes #896
Update ANA group is in local namespace list. Fixes #894
Implement namespace masking. #893
To refuse change load balancing group of the namespace if ana-group owner is another gateway #881
Allow only one host NQN when adding host with PSK keys. Fixes #869

1.3.3

28 Oct 13:15
c5c20f0
Compare
Choose a tag to compare

#912 - error handling. fix the gw to exit with error code in case the monitor client exits on error. This will allow systemctl to automatically restart the gw.
#916 - error handling. throw an error an exit with error code, in case the ping to spdk fails. This will allow systemctl to automatically restart the gw.

1.3.2

16 Sep 16:53
69cb827
Compare
Choose a tag to compare

#867 - SPDK CPU auto detect - start the spdk with the number of cores set to min between (4, number of cores on the machine), unless the mask is explicitly set in the config
#860 - Bump SPDK version to 24.05
#851 - Append group name to subsystem NQN - fix issue #851
#847 - Do not display info messages when json format is used
#845 - Handle empty log_level values as if there was no value set fix issue #845
#841 - Revert SPDK RPC fields - fix issue #841
#838 - Display an error in case operation was longer than OMAP lock duration - fix issue #838

1.3.1

03 Sep 20:26
3966c06
Compare
Choose a tag to compare

Minor ping/keepalive tweaks #778
Do not call SPDK for namespace details in namespace commands #787 (fixing issue #449)
control/server.py: use signal TERM when stopping subprocesses #788
SPDK patch to change ana group for ns - PR #798
Add TLS-PSK support - PR #808
Revert SPDK sokcet path to /var/tmp/spdk.sock - PR #810 (fixing issue #794)
Fix output redirection for get_subsystems CLI output - PR #812 (fixing issue #800)
Change --host argument to --host-nqn in host add and del commands - PR #814 (fixing issue #735)
Allow host NQN list on host add/del CLI commands - PR #817 (fixing issue #817)
A temporary fix for SPDK startup failures in CI tests - PR # 829
grpc: allow get_subsystem concurrent to other grpc requests PR #784
Unlock OMAP on exit PR #836
Display an error in case operation was longer than OMAP lock duration PR #838
Revert SPDK RPC fields PR #841
Handle empty log_level values as if there was no value set PR #845
Do not display info messages when json format is used PR #847

1.2.17

30 Jul 15:40
887c784
Compare
Choose a tag to compare

Use only NSID as the namespace id and only NQN as subsystem id - issue #757
Set SPDK log level on startup and do not lock OMAP for SPDK log commands - - issue #771 #772
Fix the breakage in test_grpc.py - issue #738
Clean up references to enable_ha as it's now always true - issue #763
Change namespace balancing group using a direct SPDK call - issue #754
Package and publish gw cli as a py package

1.2.16

10 Jul 17:10
Compare
Choose a tag to compare

#751 - Close discovery service socket before exiting

1.2.15

27 Jun 15:12
0e30e4c
Compare
Choose a tag to compare

#699 - Implement mTLS support for the monitor gRPC server

1.2.14

20 Jun 11:30
8bc6b4d
Compare
Choose a tag to compare

#690 - Fixing issue 689 - Make sure to unlock OMAP file when we're done writing to it
#695 - Fixing issue 694 - Fix typo in log message and make sure address family is not empty
#700 - Fixing issue that discovery controller crashes if no host is allowed to access the subsystem
#703 - Adding ignore_unknown_fields to json_format.Parse() calls from the omap state file. This is to allow future updates to existing elements without breaking older gws
#709 - Fixing issue 708 - Do not try to add connection to the list if there is no active qpair
#722 - Fixing issue 721 - Add an id text for update related log messages to identify the component that does the update
#612 - fixing issue 714 - fixing allocation of invalid load balancing group when CLI calls ns create in automatic LB group allocation mode. The issue is that in some cases algorithm chooses LB group that belong to deleted GW

1.2.13

28 May 16:40
23ec2b4
Compare
Choose a tag to compare
  • Revert to use size in MiBs, not bytes, in namespace_resize gRRPC call

1.2.12

28 May 11:54
d0f45d7
Compare
Choose a tag to compare
  • spdk: fix segmentation fault in call to get qpairs
  • spdk: fix log level SPDK_ERRLOG->SPDK_NOTICELOG for rados_wait_for_latest_osdmap
  • re-enable connections list command