Mon Oct 31 16:05:34 PST 2005
- Previous message: [Slony1-commit] By cbbrowne: At start of COPY_SET, we run through all the tables to "see
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Log Message: ----------- Long cluster name causes cluster to not operate properly (ID: 1436) "In set creation I allowed slony1 to create unique indexes on a handful of unkeyed tables. When the cluster name was long, the complete name it uses for those columns was truncated. The create set script/process did not detect this condition and thus all subsequent operations on those tables failed. I also had to manually drop those columns, as the uninstall process failed as well." Changed node creation to check for this, comparing cluster name size to NAMEDATALEN. Modified Files: -------------- slony1-engine/src/slonik: slonik.c (r1.49 -> r1.50) -------------- next part -------------- Index: slonik.c =================================================================== RCS file: /usr/local/cvsroot/slony1/slony1-engine/src/slonik/slonik.c,v retrieving revision 1.49 retrieving revision 1.50 diff -Lsrc/slonik/slonik.c -Lsrc/slonik/slonik.c -u -w -r1.49 -r1.50 --- src/slonik/slonik.c +++ src/slonik/slonik.c @@ -1916,6 +1916,14 @@ adminfo->version_major, adminfo->version_minor); } +#define ROWIDBITS "_Slony-I__rowID" + + if (strlen(stmt->script->clustername) + strlen("ROWIDBITS") > NAMEDATALEN) { + printf ("Cluster name %s too long to permit creation of columns containing %s - maximum name length: %d\n", + stmt->script->clustername, ROWIDBITS, NAMEDATALEN); + return -1; + } + dstring_init(&query); /* Create the cluster namespace */
- Previous message: [Slony1-commit] By cbbrowne: At start of COPY_SET, we run through all the tables to "see
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-commit mailing list