Jennifer's archive
- - 02-28-2008
- Created test plan for creating warm standby machine and breaking warm standby machine
- Created test plan for implementing Slony-1 (and breaking Slony-1)
- Followed first steps of plans:
- Start & stop server: learned that without a WAL file containing a checkpoint, the database will not recover w/out pgresetxlog. - Start & stop server: moved WAL files to different directories, did not preserve file times, database still recovered - Start & stop server: removed last WAL file from tree, broke database server. Researched why to learn that it's all about checkpoints. - Start & stop server: copy WAL files to separate locations, recover successfully from each location (database did not have knowledge of other file locale).
- Slony database successfully set up and working between tucano and equinox.
- - Security learning required before Slony-1 could be set up. Found out we are running md5 security (with .pgpass files) but that Postgres supports SSL. Also learned that Postgres is more intimately tied up with IDs in the Linux OS than some other database platforms. User postgres on equinox is completely independent of user postgres on tucano, making identification between the two servers more difficult.
- More progress made on this test plan, please read ~jennifer/database/wal_testplans.txt