Disk buffer

[GoogleCodeExporter]

Hello,

I am currently testing evtsys and it seems to work fine.
As a enhancement, are you planning to add a disk buffer feature, to preserve 
data in case of TCP connection problems?

Thank you,
Pierluigi

Original issue reported on code.google.com by pierluig...@gmail.com on 6 May 2011 at 9:02

[GoogleCodeExporter]

I have thought of adding such a feature, but the only issue is what happens 
when the syslog server is down for an extended period of time. The disk buffer 
can easily become very large. If we set a limit to the buffer, it defeats the 
purpose of having it. One solution would be to just keep track of the last 
message successfully transmitted. However, this also poses a problem when we 
attempt to send a large number of logs across the network from a busy server.

I welcome any suggestions you may have.

Original comment by sherwin....@gmail.com on 20 May 2011 at 2:47

[GoogleCodeExporter]

You could always buffer up to a maximum configurable size. Overwrite oldest 
events. This would cause gaps in the logs, but it's better than losing all of 
them. Maybe even add an mark event to indicate that gaps are present.

Original comment by lib...@gmail.com on 20 May 2011 at 10:50

[GoogleCodeExporter]

I fully agree with comment #2, that's what I had in mind too.
Thanks

Original comment by pierluig...@gmail.com on 24 May 2011 at 12:50

[GoogleCodeExporter]

This is a really great enhancement, I'd really like to see implemented.

Agreed that it needs to be handled with care. A large number of messages could 
be heavy on the hosts disk or even fill them up (i.e. DoS).

I am in favor of a configurable disk buffer which either overwrites older 
messages or doesn't accept new messages after the size/max-msg-number is 
reached.

Original comment by rosa...@gmail.com on 1 Nov 2012 at 8:52