I was about to propose that next ... you could just try back porting that uber hbck (or the relevant portions) and run the patched ruby script ... I've done this for like online meta table merges ... but not with hbck.
So - yes, that might be your solve. Good luck! --S On Mon, Jul 2, 2012 at 4:44 PM, Bryan Beaudreault <bbeaudrea...@hubspot.com> wrote: > I'm on cdh3u2 to be precise. I don't have the fixMeta option, or any other > special fix options aside from the basic -fix. > > I'll take a look at what 0.90.6 version does when it fixes meta. Maybe I > can just follow what it does there. > > Thanks again! > > On Mon, Jul 2, 2012 at 7:36 PM, Suraj Varma <svarma...@gmail.com> wrote: > >> Are you using apache hbase 0.90.4 ... or the one from CDH3? >> Check what other hbck options you have (do you have -fixMeta?) on the >> version you are on. >> >> What the uber-hbck (part of hbase-0.90.6 and later releases) does is >> "automate" this meta table surgery to fix the issue. >> >> I will let folks who have tried other repair options respond to you. >> --S >> >> On Mon, Jul 2, 2012 at 4:24 PM, Bryan Beaudreault >> <bbeaudrea...@hubspot.com> wrote: >> > Thanks for the response. >> > >> > hbck consistently shows 7 inconsistencies. I tried -fix (a few times) >> and >> > it didn't seem to fix it. I tried -details and it does seem to print out >> > more, but nothing super usable from what I can tell. >> > >> > On Mon, Jul 2, 2012 at 6:32 PM, Suraj Varma <svarma...@gmail.com> wrote: >> > >> >> Try doing an hbase hbck to see if it reports inconsistency. And do an >> >> hbase hbck -fix to see if it fixes it for you. See >> >> http://hbase.apache.org/book.html#hbck.in.depth >> >> >> >> Note that since 0.90.4 is old, some of the documented options won't be >> >> available ... but hbase hbck -fix will be available. >> >> --S >> >> >> >> On Mon, Jul 2, 2012 at 7:34 AM, Bryan Beaudreault >> >> <bbeaudrea...@hubspot.com> wrote: >> >> > Hey all, >> >> > >> >> > I'm running hbase 0.90.4, and somehow the regions for one of my tables >> >> have >> >> > become unlinked from the table itself. The Master UI shows no regions >> >> for >> >> > the table, and attempting to call Table.getRegionsInfo() returns >> nothing. >> >> > However, I can see the regions in hdfs in their proper location and >> >> doing >> >> > a scan of the table via the hbase shell does return results. I also >> get >> >> > results if I do a Get for a specific rowkey. >> >> > >> >> > How can I get these regions back associated with the table, and any >> idea >> >> > why this may have happened? I tried grepping the hmaster logs but >> didn't >> >> > see anything relevant. >> >> > >> >> > Thanks! >> >> > >> >> > Bryan >> >> >>
