On 9/13/07, Tom Lane <tgl@sss.pgh.pa.us> wrote:
I'm curious how much the WAL-recovery aspects of this patch have been
tested, because heap_xlog_clean seems quite broken.
There are quite a few crash recovery tests that one of our QA persons
(Dharmendra Goyal) has written. I can post them if necessary. We run
these tests very regularly.
Apart from these regression crash tests, I had mostly tested by
running lot of concurrent clients on UP/SMP machines, crashing
and recovering the database. We fixed quite a few issues with
these tests. I have tried crashing in middle of UPDATEs/INSERTs/DELETEs
and VACUUM/VACUUM FULL.
You have apparently
decided to redefine the WAL record format as using one-based rather than
zero-based item offsets, which would be fine if any of the rest of the
code had been changed to agree ...
I know Heikki changed that when he did the initial refactoring, but not
sure why. May be he wanted to make it more consistent.
But I don't think its broken because we collect the offsets in one-based
format in PageRepairFragmentation, WAL log in that format and redo
the same way. Am I missing something ?
Thanks,
Pavan
--
Pavan Deolasee
EnterpriseDB
http://www.enterprisedb.com