<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<meta name="generator" content="Osso Notes">
<title></title></head>
<body>
<p>----- Original message -----
<br>> > > > If flow-control is in use and one of the destinations cannot accept
<br>> the
<br>> > > messages, the other destinations do not receive any messages either,
<br>> > > because syslog-ng stops reading the source.
<br>
<br>I misunderstood what you wrote here. The docs is correct. For a single source, throttling the source side means that neither destinations receive messages.
<br>
<br>>
<br>> > this is not true. syslog-ng stops sources individually when their
<br>> > window
<br>> is full.
<br>>
<br>> But it's a quote from
<br>> <a href="http://www.balabit.com/sites/default/files/documents/syslog-ng-ose-3.3-guides/syslog-ng-ose-v3.3-guide-admin-en.html/index.html-single.html#id555527">http://www.balabit.com/sites/default/files/documents/syslog-ng-ose-3.3-guides/syslog-ng-ose-v3.3-guide-admin-en.html/index.html-single.html#id555527</a>
<br>>
<br>>
<br>>
<br>> On Mon, Nov 5, 2012 at 9:34 AM, Balazs Scheidler <<a href="mailto:bazsi77@gmail.com">bazsi77@gmail.com</a>>
<br>> wrote:
<br>>
<br>> > **
<br>> >
<br>> > ----- Original message -----
<br>> > > Thanks for your reply
<br>> > > How can I understand when it's enough to increase things? Is there
<br>> > > any manual way to get current values of each buffer, etc?
<br>> >
<br>> > well, I tend to use loggen for performance tests, also you can query
<br>> > syslog-ng internal statistics using 'syslog-ng-ctl stats'
<br>> >
<br>> > > Also since I'm logging a lot of things I'd love to know if there are
<br>> > some
<br>> > > other ways to lose messages without seeing them in "dropped"?
<br>> >
<br>> > syslog-ng counts everything it dropped using the dropped counters for
<br>> > destinations (which is a log-fifo overflow btw)
<br>> >
<br>> > messages can be lost outside syslog-ng because of transport reasons:
<br>> > * udp shouldn't be used for anything serious.
<br>> > * connection breaks can cause message loss
<br>> >
<br>> >
<br>> > >
<br>> > > > In general, performance wise you want to increase stuff
<br>> > > > (log-fetch-limit,
<br>> > > log-iw-size, flush-lines for file destinations), memory-use and
<br>> > > reliability wise you want to decrease them.
<br>> > > > Also, you have to make sure that sum(log-iw-size) < log-fifo-size.
<br>> > > So you propose just randomly tune those params? I just don't
<br>> > > understand how should I get check if it helped.
<br>> >
<br>> > no :) random tuning would be slow to converge to the ideal values.
<br>> >
<br>> >
<br>> >
<br>> > I need to see the current state of
<br>> > > each buffer(to be able to get some statistics data) to see if it
<br>> > > helps.
<br>> >
<br>> > syslog-ng-ctl stats displays the current values of statistics as a csv
<br>> > file.
<br>> >
<br>> > also you can ask syslog-ng to measure more stats by increasing
<br>> > stats-level (at the cost of some performance)
<br>> >
<br>> > >
<br>> > > And one more specific question:
<br>> > > > If flow-control is in use and one of the destinations cannot accept
<br>> > the
<br>> > > messages, the other destinations do not receive any messages either,
<br>> > > because syslog-ng stops reading the source.
<br>> >
<br>> > this is not true. syslog-ng stops sources individually when their
<br>> > window is full.
<br>> >
<br>> > > Why there is no messages about it in syslog-ng logs? It must be
<br>> > > error, don't you think so?
<br>> > > And what if I don't have flow-control enabled?
<br>> >
<br>> >
<br>>
<br>>
<br>> --
<br>> Best regards,
<br>> Koldaev Anton
<br><br></p>
</body>
</html>