Isn't Kafka like a lot faster than HTTP?
On 25 November 2015 at 02:47, Mikael Ståldal
wrote:
> I am using AsyncAppender to wrap KafkaAppender, it works fine. Consider
> setting blocking="false" and shutdownTimeout on the AsyncAppender.
>
> I have not tried async
Yes probably. But it is still a network request to a remote server.
On Wed, Nov 25, 2015 at 4:07 PM, Matt Sicker wrote:
> Isn't Kafka like a lot faster than HTTP?
>
> On 25 November 2015 at 02:47, Mikael Ståldal
> wrote:
>
> > I am using
Would using an async logger combine well with an HTTP appender like that?
Because making a REST API for this sort of thing really sounds like a
bottleneck.
On 24 November 2015 at 02:42, Mikael Ståldal
wrote:
> It is good if you can have one component less in your
It is good if you can have one component less in your system.
On Mon, Nov 23, 2015 at 8:07 PM, David KOCH wrote:
> There exists a full-fledged Java API for Elasticsearch. Still, I am not
> sure it's the best idea to lump all the intricacies of efficient indexing
> into ES into
There exists a full-fledged Java API for Elasticsearch. Still, I am not
sure it's the best idea to lump all the intricacies of efficient indexing
into ES into a log4j2 appender. Logstash does this well - there is a reason
it is part of the Elastic family.
/David
On 23 November 2015 at 18:32,
ElasticSearch has a REST API for adding data which seems to be a rather
inefficient way to implement an appender.
On 19 November 2015 at 13:36, Matt Sicker wrote:
> I just spoke with my friend who's pretty familiar with both Solr and
> ElasticSearch, and he points out that
I just spoke with my friend who's pretty familiar with both Solr and
ElasticSearch, and he points out that Solr usage might be a bit more
convoluted due to how rigidly defined search indexes are compared to ES.
It'd be like the difference between SQL and NoSQL.
On 19 November 2015 at 12:28, Ralph
The only problem I have is with the number of integrations we could do.
ElasticSearch and Solr are similar, and Solr is an Apache project, so I would
expect we would want to support both if we support either.
Ralph
> On Nov 19, 2015, at 10:07 AM, Gary Gregory wrote:
>
Is is even necessary to add Logstash to the mix? I think that Log4j 2
should be able to whatever is necessary by itself without either Flume or
Logstash.
On Wed, Nov 18, 2015 at 10:56 PM, David KOCH wrote:
> We use log4j2's SyslogAppender and a logstash UDP input source,
Amazon has a pretty rich ecosystem of stuff. Should this be in a new
log4j-amazon module or in the core module? Does it fit in no-sql?
Gary
On Nov 19, 2015 6:29 AM, "Matt Sicker" wrote:
> An ElasticSearchAppender does sound useful. We use Log4j2 in all our
> projects, so we
ElasticSearch is not Amazon specific (even though they provide it as a
service), so I don't think it should be in an amazon module. You can run
ElasticSearch yourself without Amazon.
At my previous work, we first used Log4j 2 and Graylog, and Logstash in
between. But then I developed GelfLayout (now part of log4j-core) and we
could configure Log4j 2 to send log events directly to Graylog, and
Logstash became unnecessary.
Can we do something similar with ELK (so that it become
An ElasticSearchAppender does sound useful. We use Log4j2 in all our
projects, so we don't really need to add middleware unless absolutely
necessary.
On 19 November 2015 at 07:35, Mikael Ståldal
wrote:
> At my previous work, we first used Log4j 2 and Graylog, and
We use log4j2's SyslogAppender and a logstash UDP input source, works ok
for us, no need to add Flume to the mix.
/David
On 18 November 2015 at 22:35, Matt Sicker wrote:
> Any experience with this? I was thinking of using Flume and the Flume
> appender to route all my log
14 matches
Mail list logo