46a031695c
supports rfc3164
...
also, align hostname to docs -> host
2019-03-24 20:28:03 +01:00
aeceda5caa
circolog.Message: for readers
...
refs #14
2019-03-24 19:55:01 +01:00
98a06659cb
Merge branch 'master' into statusctl
2019-01-03 12:23:20 +01:00
d1223fc170
Changing StatusResponse.Len to .Size
2019-01-03 11:51:14 +01:00
9427fe91b1
Adding ctl subcommand to print status
2019-01-03 11:15:30 +01:00
658a4bbb1e
filtering can be disabled with -tags nofilter
...
it will make your binaries way smaller
2018-12-26 01:29:39 +01:00
14e97dd43e
server-side filtering
...
circologctl filter lets you load filters on circologd
2018-12-25 03:52:53 +01:00
20269bf94e
get status
2018-12-25 01:43:40 +01:00
518b8a5588
pause: minor cleanups
2018-12-25 01:43:40 +01:00
a2303c1e1d
Implement autotoggling after timeout
2018-12-24 18:41:06 +01:00
8865335515
answer to the right command
2018-12-24 15:54:22 +01:00
dfe1e60146
Typo in comment
2018-12-19 17:30:04 +01:00
7c17c971a0
command responses
2018-11-11 19:10:53 +01:00
3f216f12f8
USR1 on circologd pause/resume
2018-11-10 18:22:23 +01:00
a990422953
HUP on circologd clears the buffer
2018-11-10 18:22:19 +01:00
f4e6370abd
make tail with param ?l= (HTTP only)
2018-11-08 19:37:03 +01:00
fbc0f96647
refactor ClientOptions
2018-11-08 19:25:40 +01:00
5ae6078805
[http] Better handling of parameters and added LastMsg to Client
2018-11-08 12:39:41 +01:00
34593d380a
refactor: hub reorganized a little
2018-08-23 12:25:07 +02:00
8568280dd1
FIX timeout issues between channels
...
we introduced non-blocking writes; however, being non-blocking when the
goroutine has no buffer means that many messages will be considered
lost.
This commit change from "non blocking" to "max 500ms; first timeout
means stop it all"; it also put a little buffer on client messages.
2018-08-23 02:05:50 +02:00
bf145240c2
proper cmd directory
2018-08-23 01:21:53 +02:00
97743eaad5
Hub resists to Client failures
2018-08-23 01:14:08 +02:00
66f32d1c05
refactor: now Hub keeps everything
...
before this there was some hidden race condition because raceCondition
is not concurrent-safe, and there was some concurrent reading and
writing.
Now everything is handled safely by the Hub.
Client now have "options" which are understood by the Hub to handle
them differently.
2018-08-23 01:04:31 +02:00
b1b83f488e
Websocket to follow logs
2018-08-22 23:51:59 +02:00
