mirror of
https://git.openafs.org/openafs.git
synced 2025-01-20 07:51:00 +00:00
68 lines
2.7 KiB
Groff
68 lines
2.7 KiB
Groff
OpenAFS Release Notes - Version 1.2.2
|
|
_________________________________________________________________
|
|
Linux clients: Major bugfix
|
|
All other systems: Minor updates
|
|
_________________________________________________________________
|
|
|
|
All systems:
|
|
- com_err patch included in 1.2.1 was missing a macro in generated
|
|
headers from .et source files. compile_et is now fixed.
|
|
- source cleanup has been undertaken to remove unused variables. This work
|
|
is part of an ongoing effort.
|
|
- fileserver will now correctly reap deleted clients. A bug previously
|
|
could preclude a client from being reaped if the client was marked
|
|
deleted and became active again before the 5 minute check LWP cleaned
|
|
it up.
|
|
- fileserver now supports InlineBulk RPC to cut down round trips on
|
|
directories including unreadable files
|
|
- fileserver is careful to avoid null pointer dereference if a callback
|
|
entry in the chain has been invalidated (possibly related to problem
|
|
fixed with reaping clients)
|
|
- bosserver logs process exit statuses
|
|
- vos can now encrypt volume dumps
|
|
- vos now has a setfields option
|
|
- vos restore now supports the -offline option to not bring volumes online
|
|
after restoring
|
|
- vos restore now supports the -readonly option to mark a restored volume
|
|
readonly. This cannot be done if a readwrite volume already exists.
|
|
- upclient cleans up after failures instead of leaving files in /tmp
|
|
- rxdebug now has a -long option to increase verbosity
|
|
- cache manager is careful to not dereference null pointers when reordering
|
|
server preferences
|
|
- bos problem where potential false errors on salvage could be reported is
|
|
fixed
|
|
- namei fileserver will attach appropriately named directories if they
|
|
contain a file named AlwaysAttach exists in them
|
|
|
|
All UNIX systems:
|
|
- dynamic root (/afs) support has been added. Invoke afsd with the -dynroot
|
|
switch.
|
|
- fix a minor vcache leak in the cache manager when a dangling mount point
|
|
is evaluated
|
|
|
|
IRIX:
|
|
- Rx now computes MTU correctly.
|
|
- Use bcopy/bzero/bcmp for kernel modules since memset/memcmp/memcpy is
|
|
not supported
|
|
|
|
Solaris:
|
|
- Solaris 9 is now supported.
|
|
- Rx now computes MTU correctly.
|
|
|
|
Linux:
|
|
- All Linux versions have a bug fixed where the kernel dentry cache would
|
|
interfere with AFS vcache entries from being deleted could lead to
|
|
an oops suggesting a VLRU cycle.
|
|
- Rx now computes MTU correctly.
|
|
- Linux 2.4 on Alpha should now build correctly
|
|
- Previously added umount_begin support has been removed as it could
|
|
cause problems for hosts where processes were still using AFS when the
|
|
cache manager was shut down.
|
|
- RC file has been cleaned up
|
|
|
|
Windows NT/2000:
|
|
- Freelance (dynamic AFS root) and AFSDB support are included
|
|
|
|
Documentation:
|
|
- NEWS file updated
|