[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: SoC: Improve syslogd
I could do a quick test. Unfortunately, it does not work. You will
always see EAGAIN - no matter if the underlying connection is broken
or not. While on the road, I did also think about this method. From a
theoretical point of view, I think it can not work. I'll elaborate a
bit later as I need to head to the next meeting (I just found time to
squeeze the quick test into today's busy schedule...).
On Thu, May 29, 2008 at 9:23 AM, Rainer Gerhards
> I thought a bit about the approach, I think there is a race condition
> in it, even if it works. Quick summary here:
> On Thu, May 29, 2008 at 9:02 AM, Rainer Gerhards
> <rgerhards%gmail.com@localhost> wrote:
>> So far just one reply, the others follow (but I need to leave soon
>> until later ;)):
>> On Thu, May 29, 2008 at 1:45 AM, Martin Schütte
>> <info%mschuette.name@localhost> wrote:
>>>> this can lead to massive data loss, even to complete data loss. Note
>>>> that this is not a direct cause of transport-tls, but of the underlying
>>>> plain TCP syslog protocol. I have more details in my blog post on the
>>>> unreliability of TCP syslog.
>>> I also thought about that message loss and I think I found a solution.
>>> Please take a look at
>>> http://mschuette.name/wp/2008/05/29/reliable-tcp-reconnect-made-easy/ and
>>> let me know what you think.
>> I think this is a nice but partial solution. Note: I did not yet prove
>> what I am talking about ;), but.. As of my understanding of TCP, this
>> does not (actually MUST not) work if the connection really breaks
>> (that is remote machine or network connection dies). In this case, you
>> do not have a real shutdown.
>> For the TLS use case, however, this should be sufficient (as we always
>> have a connection close, at least if nothing goes strangly wrong).
>> But the shutdown case should work. I'll give it a try and report back.
>> Thanks for the good idea!
Main Index |
Thread Index |