git.net

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Long GC Pauses


General best practices with Java 8:

If you have enough RAM for 24 GB heap, use G1 GC.
If you have less RAM, then use CMS with a medium-sized heap setting so the
GC time is not as long but more frequent.
Graph memory use with Grafana or something and let people know what's happening.


Your cluster:

Which version of Java?
How much RAM do your systems have?
Is it the same on all nodes?
What are your current heap settings?
Anything else?

Thanks, James Briggs.
--
Cassandra/MySQL DBA. Available in San Jose area or remote.
cass_top: https://github.com/jamesbriggs/cassandra-top



From: Rajasekhar Kommineni <rajacoast@xxxxxxxxx>
To: user@xxxxxxxxxxxxxxxxxxxx
Sent: Monday, November 19, 2018 2:33 PM
Subject: Long GC Pauses

Hi All,

My C cluster configuration.

1) 2 DC with 4 nodes each and Replication Factor of 3 per each DC
2) Writes(Bulk data load) are done to 2nd  DC and Application (reads) are done from 1st DC.
3) CMS GC

Issue : Observing long GC pauses during data load and timeouts from application (reads) during the same time.

Question :

1)Why am I seeing GC pauses on 1st DC , even though I am using stream_throughput of 16 Mb/s.
2) Is there any way to reduce the GC pause times other than changing it.

Thanks,



---------------------------------------------------------------------
For additional commands, e-mail: user-help@xxxxxxxxxxxxxxxxxxxx