/
Changelog
executable file
·90 lines (67 loc) · 4.55 KB
/
Changelog
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
Summary of versions and what has been accomplished yet...
rel-0-1: Basic infrastructure is ready. Most of the library-based utilities
work fine. VFS infrastructure is still being worked upon. Fixed some
weird threading related issues with the server side TCP based RPC
code. Finally determined that UDP/RPC with thread pools works fine,
but TCP/RPC needs a single threaded service.
rel-0-2: Finally, the VFS infrastructure is also ready and working great.
Symlinks seemed to crash the kernel module for some reason. Need to investigate
this a bit more. But basically, this should serve as the testing base against
which we need to compare the content-addressable scheme.
rel-0-3: Ok, this is a bug-fix over the previous release that just fixes
the symlink problem. Detailed message is in the CVS history log.
rel-pre[23]-freenix: This is the version that is fairly functional and usable.
Most I/O and meta-data operations seem to complete without any hiccups.
(Or so I think ;)). Checkpointed it as the snapshot at the time of the Freenix
paper submission. Fixes a lot of memory copying violations. No attempt
has been made to improve the performance and the emphasis has been on correctness.
rel-0-4: Fixed all the bugs (hopefully) after migration to an RPC-based CAS server
interface. Hurray, I can now mount a CAPFS volume from my Linux/PPC powerbook
when the data is actually stored on a x86-based machine, Seems to work pretty good
on a limited set of tests.
rel-0-4-1: Complete revamp of the build system with automagical tracking of dependance
across sub-directories etc. All namespace related functions are now prefixed
with a capfs_ instead of a pvfs_. A new configure system is also in place to
take care of platform-specific quirks. Hopefully, this should be the base
against which jason can start working on his GC code.
rel-0-4-2: More fixes for the message printing subsystem to use the LOG
macros, more beautified Makefile.in fixes for non-verbose builds. More
information can be found from the cvs commit logs. Incremental release fix.
rel-0-4-3: Cleanups in the libcas directory to get things to work despite not
setting instrumentation flag and other bugs. Build system fixes to get test
and fsck subdirectory under the purvey. A new configure option to pass in
thelocation of the MPI libraries and header files.
rel-0-4-4: More cleanups in the build process and fixed a few errors and typos
in the process.
rel-0-4-5: Added plugin support at the client daemon, and support for retrying
of failed commits.
rel-0-4-6: NEW release containing the kernel driver for 2.6.x based kernel. Needs extensive testing
though. Basic functionality was only tested in this release.
rel-0-4-6-3: Minor release that contains a few bugfixes like the 2.4 redhat
kernel workaround, 2.4 kernel makefile fixes, open RPC to honor the
fetch_hashes flags, and the number of hashes fetched at
open/gethashes/wcommit.
rel-0-5: Major release that contains tons of bugfixes, features. Most
important are vfs2.6 port merge, hcache coherence bug fixes, automatic retries
of callback registration upon manager restart, CLIENT handle caching policy
fixes, hcache update-based callbacks and resorting to that when sharer set is
size 2, truncate bug fixes, /proc/sys/capfs displays statistics on
hcache/performance related statistics.
rel-0-5-1: Minor release that should contain preliminary plugin infrastructures, performance
related fixes for reads, bumped up I/O transfer sizes, few other minor bug fixes that are described
in the cvs ci log.
rel-0-6: Quite a few bug fixes are present in this release. But most importantly it alsocomes
with a whole lot of performance statistics and tunables. This is also
incidentally the last RPC based server release for the cas server side since
that really sucks on performance.
rel-0-7: This is the first release that re-introduces the socket-based CAS server implementation
that was thrown out in favor of the RPC based implementation. It turns out that we need this to gain back our
performance loss. This has all the fixes for NO_DELAY etc but does not cache the socket fds to the
IOD servers.
rel-0-8: This is the first release that introduces a select-based implementation of the CAS
server and client-sockfd caching.
rel-0-8-1: Minor bugfix release + assorted code cleanups and print statements. Version with which
the socket-based , cached sockfd code results are being obtained.
rel-0-9-0: Major release that fixes the performance related problems with the hcache, a few critical
bug fixes affecting the Myrinet cases. Lets wait and hope that this will really help us with our
bandwidth problems :)