[relp] RELP - Handling SIGHUP correctly.

[Charles Sullivan]

rsyslogd 4.2.0, compiled with:
    FEATURE_REGEXP:                Yes
    FEATURE_LARGEFILE:            Yes
    FEATURE_NETZIP (message compression):    Yes
    GSSAPI Kerberos 5 support:        Yes
    FEATURE_DEBUG (debug build, slow code):    No
    Atomic operations supported:        Yes
    Runtime Instrumentation (slow code):    No

See http://www.rsyslog.com for more information.


On Fri, Oct 29, 2010 at 1:40 PM, <david at lang.hm> wrote:

> what version of rsyslog are you running? the behavior of syslog when it
> gets a HUP has changed drastically between versions 3 and 5.
>
> David Lang
>
> On Fri, 29 Oct 2010, Charles Sullivan wrote:
>
>  Date: Fri, 29 Oct 2010 10:58:22 -0500
>> From: Charles Sullivan <charlie.sullivan at gmail.com>
>> To: relp at lists.adiscon.com
>> Subject: [relp] RELP - Handling SIGHUP correctly.
>>
>>
>> I have a client (C) and server (S) where C is sending logs to S via the
>> RELP
>> protocol.  The server is running logrotate and sending a HUP signal to
>> rsyslog after it runs.
>>
>>   postrotate
>>                             /usr/bin/killall -HUP rsyslogd
>>   endscript
>>
>> The problem is that after the HUP signal is sent to Rsyslog, all logs sent
>> from C to S are lost for ~1 minute.  However, if I send a message from S
>> to
>> S using 'logger -p local3.info "message"' immediately after the HUP
>> signal
>> then things go through locally.  That leads me to believe that rsyslog did
>> get the HUP signal correctly and that it isn't the problem.
>>
>> Does anyone have any idea why I'd lose messages for ~1 minute?  Also, why
>> wouldn't RELP be aware of this and resend them?
>>
>> Thanks,
>>
>>


-- 
Charles Sullivan
charlie.sullivan at gmail.com