Browse Source

Merge pull request #8769 from ThomasWaldmann/update-changes-1.2

backports and changelog update (1.2-maint)
TW 1 month ago
parent
commit
3918da0ad6
2 changed files with 82 additions and 1 deletions
  1. 49 1
      docs/changes.rst
  2. 33 0
      docs/faq.rst

+ 49 - 1
docs/changes.rst

@@ -392,12 +392,60 @@ Compatibility notes:
 Change Log
 ==========
 
-Version 1.2.8 (2024-03-29)
+Version 1.2.9 (2025-04-19)
 --------------------------
 
 For upgrade and compatibility hints, please also read the section "Upgrade Notes"
 above.
 
+New features:
+
+- compact: support --dry-run (do nothing) to simplify scripting, #8300
+
+Fixes:
+
+- config: fix acceptance of storage_quota 0, #8499
+- config: reject additional_free_space < 10M (but accept 0), #6066
+- yes: deal with UnicodeDecodeError in input(), #6984
+- fix WORKAROUNDS=authenticated_no_key support for archive TAM authentication,
+  #8400
+- freebsd: fix nfs4 acl processing, #8756.
+  This issue only affected borg extract --numeric-ids when processing NFS4
+  ACLs, it didn't affect POSIX ACL processing.
+- setup.py: fix import error reporting for cythonize import, #8208
+- setup.py: detect noexec build fs issue, #8208
+- Fix paper key import error message
+
+Other changes:
+
+- allow msgpack 1.1.0
+- changed insufficiently reserved length for log message, #8152
+- docs:
+
+  - clarify excluded and included flags for dry-run, #8556
+  - update repository URLs in docs to use new syntax, #8361
+  - mount docs: apply jdchristensen's suggestion, better phrasing.
+  - FAQ: Why is backing up an unmodified FAT filesystem slow on Linux?
+  - FAQ: Why are backups slow on a Linux server that is a member of a windows domain?
+  - FAQ: add entry about pure-python msgpack warning, #8323
+  - work around sudden failure of sphinx ini lexer
+  - readthedocs theme fixes
+
+    - bring back highlighted content preview in search results.
+    - fix erroneous warning about missing javascript support.
+  - mount: add examples using :: positional argument, #8255
+  - Installation: update Arch Linux repo name
+  - recreate: remove experimental status
+- tests:
+
+  - ignore `com.apple.provenance` xattr (macOS specific)
+  - fix pytest_report_header, #8232
+- github CI: fix PKG_CONFIG_PATH for openssl 3.0
+
+
+Version 1.2.8 (2024-03-29)
+--------------------------
+
 Fixes:
 
 - check: fix return code and log level for index entry value discrepancies

+ 33 - 0
docs/faq.rst

@@ -489,6 +489,13 @@ using ``nfs=nostale_ro``. Doing so implies mounting the filesystem read-only.
 Another option is to not consider inode numbers in the files cache by passing
 ``--files-cache=ctime,size``.
 
+Why are backups slow on a Linux server that is a member of a Windows domain?
+----------------------------------------------------------------------------
+
+If a Linux server is a member of a Windows domain, username to userid resolution might be 
+performed via ``winbind`` without caching, which can slow down backups significantly. 
+You can use e.g. ``nscd`` to add caching and improve the speed.
+
 Security
 ########
 
@@ -1045,6 +1052,32 @@ A workaround is to set the option ``--files-cache=ctime,size`` to exclude the in
 number comparison from the files cache check so that files with different inode
 numbers won't be treated as modified.
 
+Using a pure-python msgpack! This will result in lower performance.
+-------------------------------------------------------------------
+
+borg uses `msgpack` to serialize/deserialize data.
+
+`msgpack` has 2 implementations:
+
+- a fast one (C code compiled into a platform specific binary), and
+- a slow pure-python one.
+
+The slow one is used if it can't successfully import the fast one.
+
+If you use the pyinstaller-made borg "fat binary" which we offer on github
+releases, it could be that you downloaded a binary that does not match the
+(g)libc on your system.
+
+Binaries made for an older glibc than the one you have on your system usually
+just work, but the opposite is not necessarily the case and can lead to misc.
+issues - like failing to load the fast msgpack code or not working at all.
+
+So: try a binary made for an older glibc.
+
+If you see this without using a "fat binary" from us, it usually means that
+msgpack is not built / installed correctly. It could be also that the platform
+is not fully supported (so the python code works, but there is no fast binary
+code).
 
 Is there a way to limit bandwidth with Borg?
 --------------------------------------------