Latest Contributions by VinceAngelo

If you enable tracing (with SDETRACELOC/SDETRACEMODE variables), you'll know what's wrong for sure, but 92sp3 != 92sp6, so your PATH order in resolving SDE.DLL could be the deciding factor. - V ... View more

It looks like you have an extra space in the ENVS path ("\ LIB", not "\LIB"). - V ... View more

You should contact Tech Support to find why the app is failing (this is the other side of a Network I/O (-10) error, so you'll need to tell them what you were doing when the gsrvr failed). - V ... View more

Are you running the supported SQL-Server 2005 build?  SP2 and SP3 are certified. - V ... View more

User schema geodatabases do not create another SDE schema.  They create parallel ArcSDE tables in a different schema, which work in concert with (but apart from) the existing SDE schema's tables. - V ... View more

The SRID model does not support a delete operation. I can't imagine how deleting unused entries could hurt (if they are truly unused, of course), but I can't imagine how leaving them in place would hurt either. - V ... View more

Creating two replicas in one database instance would be possible with any of the other DBMSes supported with ArcSDE, but Oracle uses a differrent container model and cannot be used in this manner.  If cost is the issue, you should look at using open-source PostgreSQL instead. - V ... View more

No, this is not possible.  Multiple repositories require muliple Oracle instances. - V ... View more

Is your database 10g or 11g, or did you have this problem with both databases? I rarely use 'sdelayer' to create geometry columns. I wonder if the problem is more with 'sdelayer' than the underlying SE_layer_create API call. I'm also wondering if it has to do with the lack of coordinate reference information (no "-G" and "-x" [or "-R"]). What happens if you define a proper coordref? The 'sdelayer' command accepts a flag to register the rowid column during '-o add' -- what happens if you skip the 'sdetable -o alter_reg' and just add "-C objectid" to 'sdelayer -o add'? The objectid column is required to be NOT NULL -- what happens if you define it that way with 'sdetable -o create'? - V ... View more

A search with "ORA-01480" turned up this bug report.  I avoided Oracle 10.2.0.4 and have moved on to 11g (11.1.0.7), and I haven't seen this problem. - V ... View more

Does the table exist in the database? Do references to the table exist in sde.SDE_LAYERS / sde.SDE_GEOMETRY_COLUMNS / sde.SDE_TABLE_REGISTRY / sde.SDE_COLUMN_REGISTRY ? Does a reference to the table exist in sde.GDB_OBJECTCLASSES? Do any references to the (former) objectclassid from gdb_objectclasses exist in any of the other GDB tables? Technically, Tech Support should be the proper source for direction to perform otherwise unsupported edits to geodatabase metadata, but if you are comfortable working without their direction (and have a valid current backup), you can try to clean up the metadata references yourself. Be sure to use the simplest corrective means possble (e.g., if the objectclass is missing, but SDE_* table references exist, use 'sdetable -o delete', not DROP TABLE [which would then break the SDE registry]). I did mention having a good backup, right? Be careful anyway, since a reflexive response of "Y" to an "Are you sure?" could cause even more work -- thus speaks experience! 🙂 - V ... View more

There isn't much of an issue using this for a simple feature class, but if you use versioning or complex behaviors in feature datasets you may run in issues.  ArcSDE used to use simple sequences, but the performance was sufficiently slow with bulk loading that the current "sequence and pipe" mechanism replaced it. - V ... View more

You can't do a partial upgrade.  You need to upgrade all schemas, SDE first, before extracting into CPUed server. - V ... View more

ArcSDE has had a column registry for a while now, which captures the column types when the table is registered.  Oracle is the only database which permits this sort of change, so there is no generic business process to support it.  The supported process involves creating a temporary column, populating it, dropping the old column, then adding the old name with the new type, copying from the temp column, and dropping the temp column (or just creating a new table with the new datatype and copying everything, which is far less destructive to table disk organization). IF your table is not versioned, and IF you're willing to take an unsupported (and medium risk) step you can try altering the column registry.  I generally avoid this kind of thing, even though I'm unlikely to be burned by it.  Be sure to change all the relevant properties -- the SDE_TYPE values are published in both the Client SDK and online documentation (SE_FLOAT32_TYPE is 3) -- and be sure to limit your UPDATE statement to a single row. - V ... View more

The three user schemas are all bound to a single instance; you need to move everything. - V ... View more