forked from twitter-archive/kestrel
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathChangeLog
59 lines (49 loc) · 2.08 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
2.1.5
-----
release: 12 January 2012
- Only sync to disk when there's data to be written. [Stephan Zuercher]
- Track latency measurements better, and track the percentage of java heap
reserved for in-memory queues. [Stephan Zuercher]
- Fix the startup script to work in "dash". [Matt Parlane]
- Send the correct reponse back for the memcached protocol for delete. [Matt
Erkkila]
2.1.4
-----
release: 21 November 2011
- Separate timers for journal fsync operations from those used for request
timeouts and queue expiry
- Additional statistics include set/get latency in microseconds and actual
request timeout length in milliseconds
- Upgrade to latest Netty, Ostrich and Util libraries (fixes for Netty timer
bugs)
- Creating a queue with an illegal name causes an error
2.1.3
-----
release: 13 October 2011
- Fix a bug where timers could stick around even after a client disconnected
or received an object.
- Add 2 missing server stats to the ostrich port.
2.1.2
-----
release: 29 September 2011
- Fix bug 71: connection counts could go negative. [Jeff Clites]
- Allow a queue name parameter to "dump_stats". [Jeff Clites]
- Fix potential file descriptor leak when rotating journals. [Ross Cohen]
2.1.1
-----
release: 23 September 2011
- Remove the extra linefeed from memcache "version" and "dump_stats"
repsonses. [prcm]
- Reduce the number of Promise objects created when a client isn't waiting
for one.
- Use a separate background thread for expiring queue items.
- Several improvements to the load tests. [Eric Lindvall, Jonathan Reichhold]
- Add "version" support to the text protocol. [Cory G Watson]
- Report latency of gets (hits vs misses) and sets.
- Fix a deadlock with transactions that abort and hand off the aborted item
to a waiting client that has also disconnected. [Jeff Clites]
- Don't fsync if nothing has been written to the journal since last time.
[Jeff Clites]
- Be more careful about recovering from corrupt/truncated journals, and exit
cleanly if something goes awry during startup. [Jeff Clites]
- Delete counters & gauges for queues that have been deleted. [Cory G Watson]