BIND 10 #420: Unresponsive process can block msgq
BIND 10 Development
do-not-reply at isc.org
Mon Dec 6 13:46:50 UTC 2010
#420: Unresponsive process can block msgq
---------------------+------------------------------------------------------
Reporter: shane | Owner:
Type: defect | Status: new
Priority: major | Milestone:
Component: msgq | Keywords:
Sensitive: 0 | Estimatedhours: 0
Hours: 0.5 | Billable: 1
Totalhours: 0 | Internal: 0
---------------------+------------------------------------------------------
A process that does not read it's communication channel can cause the msgq
to block. This is bad for the entire system.
The recent example is:
{{{
send(10, "\0\0\0\227\0\206{\"instance\": \"*\", \"group\": \"Xfrout\",
\"seq\": 172, \"from\": \"4cf7bb38_6 at madras\", \"to\":
\"4cf7bb38_7 at madras\", \"reply\": 336, \"type\": \"send\"}{\"result\":
[0]}"..., 155, 0
}}}
We should change the msgq to send in non-blocking mode.
What happens when msgq attempts to send and it is not able to is an open
question. One possibility is to simply close the connection. Another
possibility is to buffer the data, although we need to eventually give up
and close the connection.
--
Ticket URL: <http://bind10.isc.org/ticket/420>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list