SET zero_damaged_pages = on; VACUUM FULL location; WARNING: invalid page header in block 5869177 of relation base/17291/17420; zeroing out page REINDEX TABLE location; WARNING: invalid page header in block 5869177 of relation base/17291/17420; zeroing out page
but this blocked the complete table and took too long time (the server is productive) so I stopped this with Ctrl-C (and the "zeroing out page" was then not done).
But now I can backup the DB again:
pg_dump custdb > custdb-2013-03-18.dump pg_dump: WARNING: invalid page header in block 5869177 of relation base/17291/17420; zeroing out page
works fine (and probably has some data lost, which I don't know yet)
What I wonder is that the WARNING comes again all the time I would have expected once the block is zero out, the warning diappears.
regards Marcel
Am 18.03.2013 12:11, schrieb Marcel Ruff:
Hi,
I use psql (PostgreSQL) 9.2.2 postgresql-9.2-1002.jdbc4.jar java version "1.7.0_17" 64-Bit Server VM
on Linux 3.4.28-2.20-default
with SSD Intel 520 Series SSDs smartctl -a /dev/sdc1 does not show me errors (afaik)
and get this error: org.postgresql.util.PSQLException: ERROR: invalid page header in block 5869177 of relation base/17291/17420
I believe it is a kernel or HD problem and my idea was to dump the data and install it on another HD:
pg_dump custdb > custdb-2013-03-18.dump pg_dump: Dumping the contents of table "location" failed: PQgetResult() failed. pg_dump: Error message from server: ERROR: invalid page header in block 5869177 of relation base/17291/17420 pg_dump: The command was: COPY public.location (xxx) TO stdout;
I looks like that the dump stops. What can I do to convince the dump to continue? After migrating to another HD I could try to reimport lost data from log files etc.
What is the reason for this error? What options do I have?