Thanks for the info guys, it was all very helpful.
I do have the mslab option enabled, but it looks like this was just a gc
triggered by the load of log splitting as Stack points out. Glad to see
distributed log splitting in 0.92.
Garrett
On Thu, Apr 28, 2011 at 9:48 PM, Garrett Wu wrote:
> 2. The master started splitting logs for the hadoop11 region.
Splitting logs is a pretty intense operation. It could have run up
your master heap and CPU such that it brought on a long GC pause.
Look for a big gap in your logging just before
+1 on Todd's GC stuff. I just implemented it and it looks like there's a bit of
improvement. I'm still early on in my testing...
Sent from a remote device. Please excuse any typos...
Mike Segel
On Apr 29, 2011, at 1:35 AM, Ted Dunning wrote:
> Swap and gc are the usual culprits for this.
>
Swap and gc are the usual culprits for this.
Are you running a recent enough version to have Todd's wondrous mslab
option?
On Thu, Apr 28, 2011 at 9:48 PM, Garrett Wu wrote:
> Some snippets from the logs are pasted below. Does anyone know what may
> have caused this? Was the hang really a gar
Our HBase master and region servers all came down during a quiet period
today (no read or write traffic). After inspecting the logs, I think I've
pieced together what might have happened:
1. One of the region servers (hadoop11) timed out and started closing
it's regions and aborting.
2.
