Re: ISAM Error -136

Rob Burba wrote: > > Version 7.31.UC4 > Sun 5000 > Solaris 2.6 > > Error indicates that engine is having problems creating new extent on table. > > Table is fragmented across 8 dbspaces and has 1-3 extents per dbspace > > Has anybody else run into this problem? There was a thread on this a week or two ago look back in the IIUG Archives of CDI. Basically a table with many variable length columns (VARCHAR, TEXT, etc) can have fewer extents than a table with fewer such columns or none since the extent records and descriptions of variable length columns must share the same fixed location on disk. Art S. Kagel

OK, Here is what I have found so far. The table extents seem to be working fine; however, the problem may be with the index. The table is fragmented across 8 dbspaces, but the index is detached in a single dbspace. Here is what I find in the sysmaster:sysextents table for the index dbsname tabname start size dbsname prim_constr 2097205 1047776 dbsname prim_constr 3145731 1047776 dbsname prim_constr 11534339 1047776 dbsname prim_constr 12582915 1047776 dbsname prim_constr 31457283 1047776 dbsname prim_constr 32505859 1047776 dbsname prim_constr 33554435 1047776 dbsname prim_constr 34603011 1047776 dbsname prim_constr 35651587 1047776 dbsname prim_constr 36700163 1047776 dbsname prim_constr 37748739 1047776 dbsname prim_constr 38797315 1047776 dbsname prim_constr 39845891 1047776 dbsname prim_constr 40894467 1047776 dbsname prim_constr 41943043 1047776 dbsname prim_constr 42991619 1015033 dbsname prim_constr 44040195 45542 I have isolated a record that fails the insert and it always fails with a -271 informix error and ISAM error -136. We think we may have hit the maximum pages per fragment (which is a possible cause of the error), but I'm not sure how to verify that. Art S. Kagel <kagel@bloomberg.net> wrote in message news:392C435D.6E3B0508@bloomberg.net... > Rob Burba wrote: > > > > Version 7.31.UC4 > > Sun 5000 > > Solaris 2.6 > > > > Error indicates that engine is having problems creating new extent on table. > > > > Table is fragmented across 8 dbspaces and has 1-3 extents per dbspace > > > > Has anybody else run into this problem? > > There was a thread on this a week or two ago look back in the IIUG Archives > of CDI. Basically a table with many variable length columns (VARCHAR, TEXT, > etc) can have fewer extents than a table with fewer such columns or none > since the extent records and descriptions of variable length columns must > share the same fixed location on disk. > > Art S. Kagel >

Found the problem. It was that the index could not create a new extent because of the 16MM page limit referenced in the SERVERS_7.3 release notes. It wasn't clear to us that this would apply to a non-fragmented index associated with a fragmented table. After fragmenting the index, the problem is resolved. Rob Burba <bburba@earthlink.net> wrote in message news:TSYW4.29508$T41.713545@newsread1.prod.itd.earthlink.net... > OK, Here is what I have found so far. > > The table extents seem to be working fine; however, the problem may be with > the index. The table is fragmented across 8 dbspaces, but the index is > detached in a single dbspace. Here is what I find in the > sysmaster:sysextents table for the index > > dbsname tabname start size > > dbsname prim_constr 2097205 1047776 > dbsname prim_constr 3145731 1047776 > dbsname prim_constr 11534339 1047776 > dbsname prim_constr 12582915 1047776 > dbsname prim_constr 31457283 1047776 > dbsname prim_constr 32505859 1047776 > dbsname prim_constr 33554435 1047776 > dbsname prim_constr 34603011 1047776 > dbsname prim_constr 35651587 1047776 > dbsname prim_constr 36700163 1047776 > dbsname prim_constr 37748739 1047776 > dbsname prim_constr 38797315 1047776 > dbsname prim_constr 39845891 1047776 > dbsname prim_constr 40894467 1047776 > dbsname prim_constr 41943043 1047776 > dbsname prim_constr 42991619 1015033 > dbsname prim_constr 44040195 45542 > > I have isolated a record that fails the insert and it always fails with > a -271 informix error and ISAM error -136. We think we may have hit the > maximum pages per fragment (which is a possible cause of the error), but I'm > not sure how to verify that. > > Art S. Kagel <kagel@bloomberg.net> wrote in message > news:392C435D.6E3B0508@bloomberg.net... > > Rob Burba wrote: > > > > > > Version 7.31.UC4 > > > Sun 5000 > > > Solaris 2.6 > > > > > > Error indicates that engine is having problems creating new extent on > table. > > > > > > Table is fragmented across 8 dbspaces and has 1-3 extents per dbspace > > > > > > Has anybody else run into this problem? > > > > There was a thread on this a week or two ago look back in the IIUG > Archives > > of CDI. Basically a table with many variable length columns (VARCHAR, > TEXT, > > etc) can have fewer extents than a table with fewer such columns or none > > since the extent records and descriptions of variable length columns must > > share the same fixed location on disk. > > > > Art S. Kagel > > > > >