/
ChangeLog
96 lines (56 loc) · 3.2 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
88
89
90
91
92
93
94
95
96
2011.04.13 v 0.1.2
* Don't set umask
2010.12.23 v 0.1.0
* now only supports node versions 0.3.2 and up
* added a timeout of 30 seconds for worker death really happening
2010.12.22 v 0.0.38
* disable PING/PONG debug code
2010.11.09 v0.0.37
* Removed pipe leaks between master and workers caused by constant pinging
* Added test_pipe_leaks test
2010.11.03 v0.0.36
* Removed support for daemonizing
2010.10.27 v0.0.35
* Lots of fixes to daemonizing. Still does not work in MacOS, but I think this is not a big problem for now, since MacOS is not really used for production, just development.
2010.10.22 v0.0.34
* Added test for unix socket server.
* Corrected unix socket server argument shifting.
* Tested setuid inside test-sudo Makefile target.
2010.10.20 v0.0.32
* JSLint output is cleaner now
2010.10.20 v0.0.31
* Kill workers if they aren't dead after 30 seconds.
* New option worker_kill_timeout - when master is stopping, time to let workers kill themselves before REALLY killing them (miliseconds). defaults to 30000 (30 seconds).
* Throws Error instances instead of strings when something goes wrong.
2010.10.11 v0.0.30
* Protected worker killing from throwing an exception when the worker is no longer there, in order to not compromise the whole shutdown process.
* Fixed some tests on linux. For some reason, linux has more process affinity and it's harder to make him balance between workers.
2010.10.11 v0.0.29
* fixed connection counting on the workers
2010.10.11 v0.0.28
* fixed a racing condition while shutting down the workers
2010.10.11 v0.0.27
* fixed a possible racing condition for access to master pid file when respawning
2010.10.11 v0.0.26
* Now uses carrier (http://github.com/pgte/carrier) as a way to implement line reading on protocol
2010.10.06 v0.0.25
* Fixed race condition between launching workers and asynchronous binding on master socket
* On sudden master death, somehow the workers should eventually realize that their master has died and kill themselves.
* added fugue.workerPids()
* added options.worker_to_master_ping_interval
* Worker notifies master that he is also listening, so master can more accurately count him in in order to invoke the options.started callback.
2010.10.05 v0.0.22
* Now workers file descriptors (stdin, stdout, stderr) are the same as the master process, so there is no longer the need to pipe them from worker into master.
2010.10.05 v0.0.20
* Fixed some important leaks that were preventing node to end after fugue.stop();
* Added "started" callback to fugue.start();
2010.10.03 v0.0.19
* Reviewed the code that synchronously tries to fetch the file descriptor from the original master into the new respawned master (luismreis)
* improved fugue.stop (pgte)
* added test-fail target to test if tests are failing correctly on spawned test children processes (pgte)
* reviewed teardown of some tests (pgte)
2010.10.01 v0.0.15
* Added tests: zero_workers, one_worker, i_can_reach_many_workers, app_reload (pgte);
* Move verbose test out of log function and into configuration. If there's need to change logging at run time, then this code can be easily moved into a (new) function. (luismreis);
* new fugue.stop() function (pgte);
* options.verbose now defaults to false;