Re: GC Storm

THX for your reply ,Jonathan. On Sun, Jun 13, 2010 at 11:27 AM, Jonathan Ellis wrote: > On Sat, Jun 12, 2010 at 7:46 PM, Anty wrote: > > Hi:ALL > > I have 10 nodes cluster ,after inserting many records into the cluster, i > > compact each node by nodetool compact. > > 5000 uncompacted sstables

Re: GC Storm

On Sat, Jun 12, 2010 at 7:46 PM, Anty wrote: > Hi:ALL > I have 10 nodes cluster ,after inserting many records into the cluster, i > compact each node by nodetool compact. 5000 uncompacted sstables is unusual. did you disable compaction during your inserts? that is dangerous. > during the compa

GC Storm

Hi:ALL I have 10 nodes cluster ,after inserting many records into the cluster, i compact each node by nodetool compact. during the compaciton process ,something wrong with one of the 10 nodes , when the size of the compacted temp file rech nearly 100GB( before compaction ,the size is ~240G) the GC

Beginner Assumptions

sandra and would like to check if my assumptions are correct so far. So if someone with operational experience could confirm these I'd really appreciate it. Basically the structure I'm going for right now looks like this: One CF with LongType Keys which represent a day (eg. 20100612,

Re: Pelops - a new Java client library paradigm

Thanks for clarifying! On 13 June 2010 09:03, Miguel Verde wrote: > afaik, Cassandra does nothing to guarantee connection-level read your own > writes consistency beyond its usual consistency levels. See > https://issues.apache.org/jira/browse/CASSANDRA-876 and the earlier >

Re: Pelops - a new Java client library paradigm

afaik, Cassandra does nothing to guarantee connection-level read your own writes consistency beyond its usual consistency levels. See https://issues.apache.org/jira/browse/CASSANDRA-876 and the earlier http://issues.apache.org/jira/browse/CASSANDRA-132 On Jun 12, 2010, at 5:48 PM, Dan Washu

Re: Pelops - a new Java client library paradigm

Very nice! You mention that the connections are handled internally by Pelops, does that mean that potentially a different connection is used for each operation performed? I had assumed using the same connection for several operations with ConsistencyLevel.ONE would provide a basic level of atomic

Re: Pelops - a new Java client library paradigm

Nice going, Dominic, having a clear API for cassandra is a big step forward :) Interestingly, at hector we came up with similar approach, just didn't find the time for code that, as production systems keep me busy at nights as well... We started with the implementation of BatchMutation, but the res

Re: cassandra crashed

These look more like symptoms of a crash, than causes of it. I.e., these are the kind of messages you would see if you kill -9 a node that is in the middle of replying to another. On Fri, Jun 11, 2010 at 6:47 AM, hive13 Wong wrote: > One of our cassandra nodes suddenly crashed, then the other 2.

Re: file corruption since 0.6.2

You're getting errors from BitSetSerializer... that hasn't changed at all from 0.6.1 to 0.6.2. (or even from 0.5 to 0.6.2...) Sounds more like bad RAM to me. 2010/6/11 Lu Ming : > Many files are corrupted when our cassandra is update to 0.6.2 > COMPACTION-POOL is down caused by the following err