Gitweb: https://sourceware.org/git/?p=lvm2.git;a=commitdiff;h=6c8c8b1b244fc6a6a27c4b... Commit: 6c8c8b1b244fc6a6a27c4b3888844ad8a4750394 Parent: 0000000000000000000000000000000000000000 Author: Alasdair G Kergon agk@redhat.com AuthorDate: 2017-12-18 20:36 +0000 Committer: Alasdair G Kergon agk@redhat.com CommitterDate: 2017-12-18 20:36 +0000
annotated tag: v2_02_177 has been created at 6c8c8b1b244fc6a6a27c4b3888844ad8a4750394 (tag) tagging a1f7a483253b11e52ad5460b28cf0c75dba26659 (commit) replaces v2_02_176
Release 2.02.177.
A wide-ranging bug fix and tidy-up release, covering in particular raid, pvmove, activation and also switching to aligned 4k blocks when writing out on-disk volume group metadata.
124 files changed, 3119 insertions(+), 2485 deletions(-)
Alasdair G Kergon (31): post-release persistent filter: Skip import before rescan log: Improve various device-related messages metadata: Remove unused vg.cft_precommitted metadata: Eliminate redundant nested VG metadata log: Add io debug class regenerate man and conf lvmdiskscan: Remove unnecessary read. device: log debug when I/O bounce buffer used label: Remove unused verify functions. device: categorise block i/o device: Skip read-modify-write if replacing whole block. device: Tag I/O for each mda on a device separately in log messages. lvmetad: drop stray underscore format_text: Use explicit alignment in wrapping calc. format_text: Replace PRI with FMT. format_text: Log additional circular buffer information. format_text: Move metadata size checking into separate fn. format_text: Use absolute alignment to calculate metadata usage format_text: Adjust metadata alignment calculation. device: Internal error if writing 0 bytes to dev. metadata: Consistently skip metadata areas that failed. device: Fix an unpaired device close. format_text: Refactor and document metadata offset calculation. format_text: More refactoring of metadata offset calcs format_text: Change metadata alignment from 512 to 4096. format_text: Round size written up to multiple of 4096. format_text: Supply mempool directly to raw_read_mda_header. format_text: Separate out code paths for buffer wraparound format_text: Transfer circular buf alloc to device layer. pre-release
David Teigland (11): lvmlockd: retry on other sanlock errors lvmlockd: override unknown lock manager error numbers man: lvmlockd steps for changing lock type lvconvert: prevent combining split snapshots in shared vg lvmlockd: avoid blocking in sanlock_convert lvmlockd: recognize normal errors from sanlock_convert lvmlockd: improve error message for VG lock conflict lvmlockd: fix log print lvmlockd: use pool lock for tmeta access lvmlock: return an error if lvb cannot be written lvmlockd: clear coverity complaint
Eric Ren (1): clvmd: supress ENOENT error on testing connection
Heinz Mauelshagen (17): raid: reject conversion request to striped/raid0 on 2-legged raid4/5 raid: reject message for 2-legged raid4/5 -> striped raid: correct raid6_n_6 -> raid5 convenience type test: avoid have_raid overhead test: additional avoid have_raid overhead raid: add validation checks for reshape flags raid: ignore --stripesize on raid4/5 conversion to 1 stripe test: new reshape test under io load test: disable new reshape test until target version 1.13.1 lv_manip: allow extension on --nosync raid lv test: cleanup command test: reduce pv sizes, fix comment and stripe size test: add large tests skip option test: allow to succeed in the cluster deactivate_lvs: deactivate any missing RaidLV legs test: new raid tests test: fix trap adding proper teardown to previous, new raid tests
Marian Csontos (15): test: Rename test nothing methods test: D-Bus test_nesting requires filter Update WHATS_NEW tests: Fix common_dev_ argument handling Update WHATS_NEW Update WHATS_NEW configure: applib is not required by dbusd lvmdbusd: Fix path to python3 configure: Fix deprecation warning for lvmapp test: lvmdbusd is used for process name lvmdbusd: All tools use detected python3 test: Fix condition when detecting lvmdbusd docs: Add TESTING file lvmdbusd: Make lvmdbusd executable lvm2app: Suppress deprecation warnings for our builds
Ondrej Kozina (2): dmsetup: hide integrity encryption keys on table output test: check dmsetup hides integrity encryption keys
Zdenek Kabelac (91): snapshot: prevent repeated merging coverity: avoid memleak coverity: avoid overflow_before_widen blkdeactive: use /sbin for mdamd shellcheck: fsadm cleanup cleanup: gcc const warning cleanup: update messages in lvconvert cleanup: drop unneeded headerfiles coverity: add some initilizers tests: snasphot merging lvconvert: update delaying message tests: have some space for 2nd mda tests: detect racy systemd version tests: allow to proceed futher with test raid: protect raid4 activation cleanup: constify lv parameter cleanup: remove stack tracing for ok path reload: do not take backup with suspended devices lvremove: for unused cache deactive sublv lv_lock_holder: unused cache-pool is not lock holder tests: use nonflusing status pvmove: snapshots need to go with origin device pvmove: return pvmove itself layers: collect only lock holding LVs activate_lvs: deactivate on uniterate activation: check subLV before skipping resume activation: suspend pvmove using lv. pvmove: always flush when working with pvmove tree activate_lvs: use exclusive activation cleanup: update log messages pvmove: change locking on finish pvmove: activation changes pvmove: remove unusued code tests: pvmove supports more cases suspend: handle start of pvmove resume: secure critical section locking: pvmove is locking holding LV libdm: preload propagates delayed resume mirror: use lv_update_and_reload_origin activation: automaticaly discover pvmove holders pvmove: finish no longer needs explicit resumed toollib: avoid printing default stripe size cache: use conditional in warning message cleanup: reduce couple debug lines cleanup: using max args consistently cleanup: update error messages tests: typo tests: more places where missing should not appear activation: enhance holders detection pvmove: enhance delayed_resume logic toollib: improve stripes args reading cmdline: avoid overrun on very large numbers. cleanup: reuse existing macro pvmove: move code from tools to lib pvmove: add missing segment merging activation: split priority from memory locking activation: extend resume validation activation: avoid rechecking pvmove node suspend: optimize generated list cleanup: simplify _deps code cleanup: remove debug code libdm: watch for failing _info_by_dev libdm: support for replicator target is dropped cleanup: drop unused header cleanup: futher code reduction cleanup: simplier error message cleanup: drop impossible test case cleanup: mark success at the end cleanup: stack tracing cleanup: use _node_name clenaup: use log_warn cleanup: use log_warn cleanup: use existing define with prefix cleanup: drop unused define cleanup: messsage cleanup cleanup: drop unneeded check activation: also lock memory for clustered locking thin: always clear memory before parsing status thin: missing type is error libdm: add help func _get_last_load_segment libdm: avoid checking status on activation libdm: use delay_resume_if_extended libdm: drop extra structure copy cleanup: reorder calling of pool checking tools cleanup: use log_warn debug: update debug msg vgchange: drop extra counting activation: keep priority till memlock_unlock thin: regression fix for metadata checking cleanup: switch to standard query cleanup: debug message