So those are the same two regions (region1 in the first error is region1 in the second error and so on)? If so, it's probably a parent and a daughter region and somehow (due to a bug, can't tell for sure what exactly) they were both re-deployed.
To fix, I think you could merge the two regions manually by first disabling the table, moving the files from one region to the other, deleting the folder / entry in .META. for the former region, and finally re-enable the table. One thing tho is that you have to keep the chain of keys intact meaning that the end key of region1/2 merged should be the start key of whatever comes next (let's call it region3). If you aren't sure which region to choose and you don't want to share your region names here, feel free to send me the .META. entries offline so that I can advise you. Stack has more experience than me doing that sort of fixups so he might have another idea, but since he works next to me we could figure it out together as long as we have the region keys :) Thx, J-D On Mon, May 16, 2011 at 12:38 PM, Andy Sautins <[email protected]> wrote: > > J-D, > > I was able to run the 0.90.3 rc hbck on our hbase cluster and it does give > more detail. It gives more details. Two specific errors: > > - ERROR: (regions <region1> and <region2>) Two regions have the same > startkey: > - ERROR: (regions <region1> and <region2>) There is an overlap in the > region chain. > > That is more detail, although I'm not really sure what to do with that > information. Is there a way to effectively repair from these two errors? > > Thanks > > Andy > > > -----Original Message----- > From: [email protected] [mailto:[email protected]] On Behalf Of Jean-Daniel > Cryans > Sent: Monday, May 16, 2011 1:05 PM > To: [email protected] > Subject: Re: inconsistencies reported by hbck... > > Would you be able to patch in > https://issues.apache.org/jira/browse/HBASE-3695 and see what hbck > tells you now? Else you could try using the 0.90.3 rc0 which has it > too: http://people.apache.org/~stack/hbase-0.90.3-candidate-0/ > > J-D > > On Sun, May 15, 2011 at 9:11 AM, Andy Sautins > <[email protected]> wrote: >> >> I finally got around to running 'hbase hbck' on our HBase cluster. We >> noticed the following error being reported: >> >> Chain of regions in table myTable contains less elements than are listed in >> META; visited=671, edges=681 >> ERROR: Found inconsistency in table myTable >> >> Can anyone help shed some light on what that error means and how I might >> resolve it? It does seem like the table is in somewhat of a bad state ( >> e.g., there are two regions that have blank start key but have different end >> keys ). >> >> We are running from the cloudera CDH3U0 distribution. Note also this >> table was originally created and running for a long time under the early .89 >> releases. I did run both 'hbase hbck -fix' and 'check_meta.rb -fix' and >> still have the above inconsistency so it looks like I need to take manual >> action at this point but I'm a little stumped on how to go about it. >> >> Any insight would be appreciated. >> >> Thanks >> >> Andy >> >> >> >
