Re: snapshot timeouts

2014-10-08 Thread Brian Jeltema
Thanks for the quick responses. I’ll get back on this later; I discovered that HBase didn’t restart properly after changing the timeouts, so the second ERROR may be a side-effect of that. I also just discovered that the table in question was not pre-split properly, and the region distribution is

Re: snapshot timeouts

2014-10-08 Thread Brian Jeltema
Sorry, I usually include that info. HBase version is 0.98. hbase.rpc.timeout is the default. When the ‘ERROR: Call id….’ occurred, there was no stack trace. That was the entire error output. Before I increased the snapshot timeout parameters, the timeout I was seeing looked like: ERROR: org.a

Re: snapshot timeouts

2014-10-08 Thread Jean-Marc Spaggiari
Also looking for the number of regions for this table, and the number of region servers... 2014-10-08 15:18 GMT-04:00 Ted Yu : > Can you give a bit more information : > > the release of hbase you're using > value for hbase.rpc.timeout (looks like you leave it @ default) > more of the error (pleas

Re: snapshot timeouts

2014-10-08 Thread Ted Yu
Can you give a bit more information : the release of hbase you're using value for hbase.rpc.timeout (looks like you leave it @ default) more of the error (please include stack trace if possible) Cheers On Wed, Oct 8, 2014 at 12:09 PM, Brian Jeltema < brian.jelt...@digitalenvoy.net> wrote: > I’m

snapshot timeouts

2014-10-08 Thread Brian Jeltema
I’m trying to snapshot a moderately large table (3 billion rows, but not a huge amount of data per row). Those snapshots have been timing out, so I set the following parameters to relatively large values: hbase.snapshot.master.timeoutMillis hbase.snapshot.region.timeout hbase.snap