Changelog for
xapi-clusterd-0.9.0-1.el7.centos.x86_64.rpm :
* Mon Apr 23 2018 Edwin Török
- 0.9.0-1- CA-288287: Use TLSv1.2 in xapi-clusterd- CA-288287: be explicit about crypto cipher/hash used
* Tue Apr 17 2018 Edwin Török - 0.8.0-1- CA-288227: make wakeups more robust- Convert unit tests to alcotest- CA-287878: Start & stop remote API server as needed bound to specified IP
* Tue Apr 03 2018 Edwin Török - 0.6.0-1- CP-27269: build: fix coverage rules- Fix license in opam file
* Fri Mar 02 2018 Edwin Török - 0.5.0-1- Drop dependency on xapi-client from main binaries- fix build rules: need to build/install only xapi-clusterd package- Fix make install rule to install only xapi-clusterd package- Add makefile rule for running vagrant tests explicitly called \'make realtest\'- [CA-264360] Ensure cluster node IDs are assigned from 1- CP26330: Added json flag to diagnostics dbg call- CP-26197, CA-279041: prevent race between on_startup and API calls- [CP-26201] Add exploratory behaviour tests- [CP-26201] Disallow remove when a node is disabled- Change logging to stdout instead of systemd- Reindent all files- CA-279944: Handle CS_ERR_TRY_AGAIN from cfg-tool -R- make retest: add back --failed- CA-279002: use per-host mutexes in fake test- Properly wrap exceptions in RPC- CA-279002: add a xapi-clusterd unit test for parallel disable- CA-279002: mutex around start and introduce stop_cluster_lock- CA-279002: retry when a host returns ENOENT on dlm_open_lockspace- CA-279002: do not allow remote_join/remote_eject to get executed when we\'re ejecting the local node already- make retest: rebuild the test executable before rerunning- [CP-25914] Use TLS for communication- [CA-280422] Do not start a server on local port by default- Reindent all files- CP-24697: v5 update with latest code changes- Change from port 2375 to 8895 for Vagrant tests- CA-280717: retry eject on next node when RPC fails- CA-280717: fix wrong transaction handling on parallel boot- [CP-26803] Update design to add cluster force destroy- design: update port 2375->8895- design: add basic description of configuration/interface provided by corosync/dlm- design: existing HA already assumes that Xen watchdog is reliable- design: testing and TLS verification- design: whitespace- design: signoff draft v6- CP-24697: design: signoff draft v6- design: markdown linting- CA-280717: refactor remote APIs- CA-280717, CP-26558: overhaul locking- CA-280717: ensure config version is always increasing- CA-281971: xcli destroy should not hang when clustering is not enabled- Update to latest upstream APIs for ring3 merge- fix opam dependencies- CA-283650: add unit test for self-RPC- CA-283722: Ensure we really leave the DLM lockspace before stopping DLM- CA-283722: Log timeouts- Use port 8896 for xapi-clusterd
* Thu Mar 01 2018 Edwin Török - 0.4.0-1- Fix bugtool paths- CP-26038: Change bugtool layout for installation- [CP-24916] Use --enable_concurrent_fencing dlm option- [CP-25892] (Partial) Set token and token_coefficient in corosync.conf- [CP-25892] Wire token and token_coefficient params- CA-274624: retry join after a failure- CA-276001: retry on failed startup- CP-26038: gather corosync/sbg/dlm debug output in bugtool- CP-26038: log fencing events at LOG_CRIT level- reindent- CP-26038: collect cmapctl output too- CA-273683: move remote interface back into xapi-clusterd- CP-26038: log RPC results- CP-26038: plumb through debug task- CP-26038: plumb through debug task- reindent: use newest ls-files rule- in make test only build xapi-clusterd- CP-26038: add logging for exit hooks- CP-25892: increase API timeout- CP-25890: do not disable on shutdown- CP-26157: do not raise async exceptions from cohttp- [CP-26175] Use corosync timeouts passed from xapi- [CP-26157] Add support for passing timeouts to cluster Init- [CP-25892] Use the computed timeout where tests needs to wait- Add debug info to nodes_booted assert- log exception when fistpoint fails- [CP-26157] Adjust timings for unit tests- CA-276677: 16 host test- CA-276988: do not kill already running corosync on startup- fix wrong units for token timeout in vagrant tests: it cannot be 20ms- CA-276677: workaround for kernel hang in dlm_new_lockspace- Drop dependency on xapi-client from main binaries- fix build rules: need to build/install only xapi-clusterd package- Fix make install rule to install only xapi-clusterd package- Add makefile rule for running vagrant tests explicitly called \'make realtest\'- [CA-264360] Ensure cluster node IDs are assigned from 1- CP26330: Added json flag to diagnostics dbg call- CP-26197, CA-279041: prevent race between on_startup and API calls- [CP-26201] Add exploratory behaviour tests- [CP-26201] Disallow remove when a node is disabled- Change logging to stdout instead of systemd- Reindent all files- CA-279944: Handle CS_ERR_TRY_AGAIN from cfg-tool -R- make retest: add back --failed- CA-279002: use per-host mutexes in fake test- Properly wrap exceptions in RPC- CA-279002: add a xapi-clusterd unit test for parallel disable- CA-279002: mutex around start and introduce stop_cluster_lock- CA-279002: retry when a host returns ENOENT on dlm_open_lockspace- CA-279002: do not allow remote_join/remote_eject to get executed when we\'re ejecting the local node already- make retest: rebuild the test executable before rerunning- [CP-25914] Use TLS for communication- [CA-280422] Do not start a server on local port by default- Reindent all files- CP-24697: v5 update with latest code changes- Change from port 2375 to 8895 for Vagrant tests- CA-280717: retry eject on next node when RPC fails- CA-280717: fix wrong transaction handling on parallel boot- [CP-26803] Update design to add cluster force destroy- design: update port 2375->8895- design: add basic description of configuration/interface provided by corosync/dlm- design: existing HA already assumes that Xen watchdog is reliable- design: testing and TLS verification- design: whitespace- design: signoff draft v6- CP-24697: design: signoff draft v6- design: markdown linting- CA-280717: refactor remote APIs- CA-280717, CP-26558: overhaul locking- CA-280717: ensure config version is always increasing- CA-281971: xcli destroy should not hang when clustering is not enabled- Update to latest upstream APIs for ring3 merge- fix opam dependencies- CA-283650: add unit test for self-RPC- CA-283722: Ensure we really leave the DLM lockspace before stopping DLM- CA-283722: Log timeouts
* Wed Nov 29 2017 Edwin Török - 0.2.0-1- Add some one-node tests- tests: give nodes a chance to fence- Remove is_enabled from cluster_config- no need to use Lwt_list when the predicate is a direct function- CP-24769: report on service status before changing it- Ensure we only Db.{set,remove} inside transactions- CP-24769: disable support for last node- impl_corosync: is_enabled: remove unnecessary use of Lwt_list- test/test: add TODO to look for invalid configurations when computing is_enabled- test_interface: remove Disable action- test_interface: rename Shutdown to Disable to match API- test_interface: rename Enable to Init- CP-24908: implement Local.enable API call- test_interface: add Enable action performing Local.enable- Add test scenarios involving Local.enable- test_enable: Add scenarios to test node ID stability- lab: rename node to addr- test: Include node address in node_state- test: rewrite assertRunningSanely- assertRunningSanely: remove redundant code- assertRunningSanely: check powered off nodes are not powered on- CP-24203: remote_rejoin: also update all_members when IP changes- lib/systemio: run: print stderr on failure- vagrant_ssh: allow ignoring of return code- CP-24203: test: support change of IP address- CP-24203: add tests involving IP address changes- Add scenario 6.12b- test/fake: dissociate host_db from old IP address- CA-268478: update IP address on singleton node enable- CA-268478: relax is_enabled check to ignore IP address- CA-268478: enable: update all_members, enabled_members in case of IP change- Pass nodeid to RPC function- test/fake: check whether RPCs are going to right address- Support Local.declare_changed_addrs API call- test_enable: Add more IP address change test scenarios- Print actual error when we fail to start the test- Update build dependencies to new xen-api-client-lwt- Fix test/vagrant.ml build with new xen-api-client-lwt- Scenario 3.8: add explanatory comment- Scenario 3.4: check no fencing and split into two tests- Scenario 3.5: various fixes- distribute_state: perform reload as part of distribute_state- distribute_state: first ping the hosts to check for liveness- impl_corosync: provide cluster_config to start- test/fake: make wrpc not rely on cluster_config- corosync_actions: pass the node\'s address to the rpc function- Workaround Dlm.leave not returning EBUSY even when it can\'t leave- CA-270008: leave dlm lockspace with force- CA-270008: stop corosync during Remote.eject, not before (or after)- CA-270008: distribute_state: invoke start on the caller- vagrant: allow up to 8 VMs- distribute_state: use iter_p rather than iter_s- make SBD timeout a variable- CP-24803: add timeouts to RPC calls- CP-24803: add timeout for test API calls- CP-25392: Add Scenario 1.9b- corosync_actions: remove dead type- CP-25547: Support dynamic failure insertion in fake testing- Fix xapi-clusterd dependencies- corosync_real: don\'t fail in check_fistpoint- Use Dlm_sbd.with_timeout on all Remote API calls- impl_corosync: join: don\'t start corosync after Remote.join- tests: use library functions nodes_booted, nodes_quorate where possible- tests: refactor extraction of node_id- tests: refactor node indexes- Refactor all_members out of cluster_config- Deduplicate fields from diagnostics and node_state- diagnostics: move is_enabled from node_state- diagnostics: also return live config- diagnostics: also return next_cluster_config- impl_corosync: don\'t use find_first_success in disable, leave- distribute_state: refactor call_everyone- distribute_state: return which hosts call_everyone succeeded on- CA-271869: improve failure handling during distribute_state- diagnostics: support config_valid field- Add scenario 3.4d- Add scenario 3.4e- Add scenario 3.4f- Scenario 3.4c: verify behaviour if dead node comes back- distribute_state: ping and invalidate all current members- Add scenario 3.4g: abort success test- Add scenario 3.4h: node being disabled failing before invalidate- distribute_state: catch errors in pre_reload_f and post_reload_f- Add scenario 3.4i: disabled node failing in pre-reload- Add scenarios 2.2b--h- test_client: temporarily disable live_/next_enabled_members checks- vagrant: remove dead code (Angstrom parser)- corosync_real: refactor config file paths- corosync_real: parse corosync.conf to provide diagnostics.next_cluster_config- corosync_real: parse cmap to provide diagnostics.live_cluster_config- Revert \"test_client: temporarily disable live_/next_enabled_members checks\"- CA-273755: fix fd leak in Systemio.read_file- test/test: restrict scope of open Xapi_cluster.Fake_signaller- Add config_version field to cluster_config- CP-24481: increment config_version on distribute_state- corosync_fake: implement config_version semantics- CP-24481: enable and fix scenario 4.5- Add scenario 4.5b: a smaller version of 4.5- vagrant: raise nicer error on failure to get IP- Fix timeouts for API calls that cause nested API calls- fist: remove double-quotes in Fist.to_string- fist: associate fistpoints with hostname, not address- fist: move {arm,disarm}_fistpoint from Corosync_fake to Fake_signaller- fist: minor refactoring of fistpoint code- fist: add prepare_wait_fistpoint_reached to Fist.SIGNALLER- fist: split SIGNALLER into test-side and daemon-side modules- CP-25654: support FIST points in Vagrant tests- on_startup: don\'t exit if cluster config is invalid- Reformat jbuild file in the usual \'sorted columnar dependencies\' format- CP-24803: limit retries based on maximum time, not count- Properly restore indent after failed RPC call- CP-24803: add fallback timeout on local API calls- CA-267673: Avoid stuck test due to filling stdout- CA-267673: do not retry indefinetely in enable- CA-267673: report back actual error in enable- add mtime version constraint- Add xapi-clusterd design draft (v1)- remove redundant comment- destroy, leave: also nuke any lurking invalid state- Remove some legacy references to pcs- design: add detail on testing and clustering configuration changes- TLS and firewall ports- msc: Update all message sequence charts- remove dead code: Corosync_fake.get_addr- remove dead code: Fake_signaller.forget_fistpoints- remove dead code: Impl_corosync.(>>|=)- remove dead code: Test.remove_node- rejoin_inner: only work if there are other nodes to rejoin us- Add scenario 6.5d: enable of node in 2-node cluster- Add scenario 6.15d: cannot declare IP of enabled node changed- declare_changed_addrs: fix log message- design: add note about adding more FIST points- design: always join DLM lockspace when starting corosync- Local.join: no need to write cluster config to DB- rejoin_inner: no need to start corosync- declare_dead: add comment- design: sign-off
* Wed Sep 13 2017 Edvin Török - 0.1.4-1- Update to latest master to fix build error with patch queue- CA-264366: make shutdown idempotent- CA-264363: make remove idempotent- CP-24104: Make code coverage work with xapi-clusterd-test- CA-264518: improve debugging around authkey decoding- CA-264518: make remote_rejoin idempotent- More flexibility in starting services (CP-24480, CA-265273)- CA-265457: impl_corosync: handle types of dlm_error other than ENOENT- CA-265457: test/vagrant: sync after shutdown- CA-265457: Scenario 4.3: allow longer after DeclareDead
* Fri Aug 18 2017 Mark Syms - 0.1.3-1- Run unit tests in build check step
* Thu Aug 17 2017 Jonathan Davies - 0.1.2-2- Add build-time dependencies required by unit tests
* Fri Aug 04 2017 Edvin Torok - 0.0.9-1- CP-23722: Initial packaging