This isn’t true if your clustering is time based because the read path can
selectively include/exclude sstables based on the clustering keys
--
Jeff Jirsa
> On Oct 25, 2018, at 12:26 PM, Dor Laor wrote:
>
> TWCS is good for time series but if your workload updates the same keys
> within
thank you let me review.
On Fri, Oct 26, 2018 at 3:35 PM Sebastian Estevez <
sebastian.este...@datastax.com> wrote:
> Here's the metrics you want:
> http://cassandra.apache.org/doc/latest/operating/metrics.html#table-metrics
>
> The best practice is to run fewer bigger tables. If it's a lot of
Here's the metrics you want:
http://cassandra.apache.org/doc/latest/operating/metrics.html#table-metrics
The best practice is to run fewer bigger tables. If it's a lot of tables
you're likely out of luck aside from throwing more RAM at the problem.
All the best,
Sebastián Estévez | Vanguard
anyone has any idea on this?
On Thu, Oct 25, 2018 at 11:35 AM Jai Bheemsen Rao Dhanwada <
jaibheem...@gmail.com> wrote:
> Hello,
>
> I am running into a situation where huge schema (# of CF) causing OOM
> issues to the heap. is there a way to measure how much size each column
> family uses in
Hello
Any way to temporarily make the node under maintenance invisible from
> "nodetool status" output?
>
I don't think so.
I would use a different approach like for example only warn/email when the
node is down for 30 seconds or a minute depending on how long it takes for
your nodes to
I was able to make this work using the following UDF/UDA:
CREATE OR REPLACE FUNCTION agg_set_func(state tuple>, val
set) CALLED ON NULL INPUT RETURNS tuple> LANGUAGE
java AS
$$
if (val == null) {
return state;
}
Set s = state.getSet(1, Long.class);
s.addAll(val);
I have script that parses "nodetool status" output and emails alerts if any
node is down. So, when I stop cassandra on a node for maintenance, all
nodes stats emailing alarms.
Any way to temporarily make the node under maintenance invisible from
"nodetool status" output?
Thanks
Those changes where for testing code only.
https://github.com/datastax/cpp-driver/commit/ffc9bbd8747b43ad5dcef749fe4c63ff245fcf74
The driver has compiled with fine w/ C++98 for some time now. The
encoding/decoding doesn't make any assumptions about endianess so it should
work fine on big-endian
Hello,
A client of mine has recently asked me to have a look at Datastax Opcenter’s
(nodetool and JMX?) statistics, such as time taken by read and write queries,
and to write an instruction manual on how to interpret them.
Unfortunately, said statistics are quite difficult to analyze. For
