A previous post here mentioned using db_dump and db_load to compact the
component databases of a subversion fs. Curious, I decided to test this,
with the following rather strange results:
1) Dump the strings db to a file. OK.
2) Try to load the dump into a new db file ("stringsnew"). Fails with a
large amount of scary error messages, and leaves behind a stringsnew file
which is far too small.
3) Rerun the load. It gives one warning, and then completes successfully.
4) Verify success by re-dumping and comparing with the first dumpfile.
Which really is just bizzarre.
Has anyone experienced anything like this?
Cygwin, WinXP, db-4.1.25.1, svn r6360.
1) $ db_dump strings > sdmp
2) $ db_load stringsnew < sdmp
db_load: Lock table is out of available locks
db_load: Not enough core
db_load: Lock table is out of available locks
db_load: Fatal error in abort of an allocation
db_load: PANIC: Not enough core
db_load: fatal region error detected; run recovery
db_load: PANIC: DB_RUNRECOVERY: Fatal error, run database recovery
db_load: fatal region error detected; run recovery
db_load: fatal region error detected; run recovery
db_load: DB->close: DB_RUNRECOVERY: Fatal error, run database recovery
db_load: fatal region error detected; run recovery
db_load: dbenv->close: DB_RUNRECOVERY: Fatal error, run database recovery
3) $ db_load stringsnew < sdmp
db_load: fatal region error detected; run recovery
4) $ db_dump stringsnew > sdmp2
db_dump: fatal region error detected; run recovery
$ cmp sdmp sdmp2
$
Max.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Jun 29 12:47:28 2003