Yea, it used to throw some IllegalArgumentException. Now, I guess it just gives the wrong result.
Should shell compensate or should we fix the underlying issue? It seems we could probably tweak the comparator to deal with the case when one of the comparison operands has no commas. -Todd On Mon, Jul 18, 2011 at 9:34 AM, Stack <[email protected]> wrote: > Ugh. I should have remembered this. Shell should probably compensate > when the target of scan is one of the catalog tables. Good on you > Todd, > St.Ack > > On Sun, Jul 17, 2011 at 11:22 PM, Lars George <[email protected]> > wrote: > > Wow, awesome Todd: > > > > hbase(main):004:0> scan '.META.', {STARTROW => 'testtable2', LIMIT => 1} > > ROW COLUMN+CELL > > 2011-07-17T23:20:45.480-0700: 58.760: [GC 58.760: [ParNew: > 19136K->2112K(19136K), 0.0098980 secs] 21990K->6759K(83008K), 0.0099430 > secs] [Times: user=0.05 sys=0.00, real=0.01 secs] > > TestTable,,1294648777729.1c1644f4a column=info:regioninfo, > timestamp=1294648777898, value=REGION => {NAME => > 'TestTable,,1294648777729.1c > > 1304442d9c8483a6d167d48. 1644f4a1304442d9c8483a6d167d48.', > STARTKEY => '', ENDKEY => '', ENCODED => 1c1644f4a1304442d9c8483a6d1 > > 67d48, TABLE => {{NAME => 'TestTable', > FAMILIES => [{NAME => 'col1', BLOOMFILTER => 'NONE', REPLICATIO > > N_SCOPE => '0', COMPRESSION => 'NONE', > VERSIONS => '3', TTL => '2147483647', BLOCKSIZE => '65536', IN_ > > MEMORY => 'false', BLOCKCACHE => > 'true'}]}} > > ... > > 1 row(s) in 0.3990 seconds > > > > hbase(main):005:0> scan '.META.', {STARTROW => 'testtable2,,', LIMIT => > 1} > > ROW COLUMN+CELL > > testtable2,,1310754563673.d720f2ae column=info:regioninfo, > timestamp=1310754573539, value=REGION => {NAME => > 'testtable2,,1310754563673.d > > 6a4aec4fce746f6a9c1376e3. 720f2ae6a4aec4fce746f6a9c1376e3.', > STARTKEY => '', ENDKEY => 'row-10', ENCODED => d720f2ae6a4aec4fce74 > > 6f6a9c1376e3, TABLE => {{NAME => > 'testtable2', FAMILIES => [{NAME => 'family', BLOOMFILTER => 'NONE', > > REPLICATION_SCOPE => '0', COMPRESSION > => 'NONE', VERSIONS => '3', TTL => '2147483647', BLOCKSIZE => '6 > > 5536', IN_MEMORY => 'false', > BLOCKCACHE => 'true'}]}} > > ... > > 1 row(s) in 0.0440 seconds > > > > This also makes the STOPROW work: > > > > hbase(main):006:0> scan '.META.', {STARTROW => 'testtable2,,', STOPROW => > 'testtable3'} > > ROW COLUMN+CELL > > 0 row(s) in 0.0220 seconds > > > > hbase(main):007:0> scan '.META.', {STARTROW => 'testtable2,,', STOPROW => > 'testtable3,,'} > > ROW COLUMN+CELL > > testtable2,,1310754563673.d720f2ae column=info:regioninfo, > timestamp=1310754573539, value=REGION => {NAME => > 'testtable2,,1310754563673.d > > 6a4aec4fce746f6a9c1376e3. 720f2ae6a4aec4fce746f6a9c1376e3.', > STARTKEY => '', ENDKEY => 'row-10', ENCODED => d720f2ae6a4aec4fce74 > > 6f6a9c1376e3, TABLE => {{NAME => > 'testtable2', FAMILIES => [{NAME => 'family', BLOOMFILTER => 'NONE', > > REPLICATION_SCOPE => '0', COMPRESSION > => 'NONE', VERSIONS => '3', TTL => '2147483647', BLOCKSIZE => '6 > > 5536', IN_MEMORY => 'false', > BLOCKCACHE => 'true'}]}} > > ... > > testtable2,row-10,1310754563677.81 column=info:regioninfo, > timestamp=1310754573667, value=REGION => {NAME => > 'testtable2,row-10,131075456 > > 73e0623a0f82c5b308790737900b60. > 3677.8173e0623a0f82c5b308790737900b60.', STARTKEY => 'row-10', ENDKEY => > 'row-20', ENCODED => 8173e062 > > 3a0f82c5b308790737900b60, TABLE => > {{NAME => 'testtable2', FAMILIES => [{NAME => 'family', BLOOMFILTER > > => 'NONE', REPLICATION_SCOPE => '0', > COMPRESSION => 'NONE', VERSIONS => '3', TTL => '2147483647', BLO > > CKSIZE => '65536', IN_MEMORY => > 'false', BLOCKCACHE => 'true'}]}} > > ... > > > > So it is caused by the same issue. Should be fixed, right? > > > > Lars > > > > > > On Jul 18, 2011, at 4:36 AM, Todd Lipcon wrote: > > > >> Try with STARTROW => "testtable2,,"? > >> I seem to remember some weirdness about the start/stop row having to > have > >> the meta "format" with commas, etc > >> > >> -Todd > >> > >> On Fri, Jul 15, 2011 at 2:01 PM, Stack <[email protected]> wrote: > >> > >>> On Fri, Jul 15, 2011 at 11:40 AM, Lars George <[email protected]> > >>> wrote: > >>>> Using the start row, it prints the entire META. > >>> > >>> Starting with the STARTROW. > >>> > >>>> Setting a stop row it does > >>>> not print a single row: > >>> > >>> > >>> Yeah. I see that too in 0.90.3. > >>> > >>> St.Ack > >>> > >>>> > >>>> hbase(main):003:0> scan '.META.', { STARTROW => 'testtable2', STOPROW > => > >>>> 'testtable3' } > >>>> ROW COLUMN+CELL > >>>> > >>>> 0 row(s) in 0.0220 seconds > >>>> > >>>> I have a testtable2 in this case. Am I stuffing up the syntax? > >>>> > >>>> Lars > >>>> > >>>> On Fri, Jul 15, 2011 at 8:36 PM, Stack <[email protected]> wrote: > >>>> > >>>>> Passing 'STARTROW' in 0.90.3 seems to work. What you see Lars? > >>>>> St.Ack > >>>>> > >>>>> On Fri, Jul 15, 2011 at 11:31 AM, Lars George <[email protected] > > > >>>>> wrote: > >>>>>> Hi, > >>>>>> > >>>>>> I am on 0.90.1 (CDH3u0) but noticed that on trunk as well: doing > scan > >>> of > >>>>>> .META. using a start and stop row does not work. The start row does > >>>>> nothing, > >>>>>> i.e. all is shown, and setting the stop row is the opposite, i.e. > >>> nothing > >>>>> is > >>>>>> shown. Is that borked? > >>>>>> > >>>>>> Lars > >>>>>> > >>>>> > >>>> > >>> > >> > >> > >> > >> -- > >> Todd Lipcon > >> Software Engineer, Cloudera > > > > > -- Todd Lipcon Software Engineer, Cloudera
