WAL failure?

"Jim Buttafuoco" <jim@spectrumtelecorp.com> writes:

My system crashed today (don't know why yet.) After i started the
postmaster the log file this entry in it.
DEBUG: ReadRecord: invalid magic number 0000 in logfile 4 seg 34 off
15777792
Is this a problem???

This looks like a fairly normal WAL-log-recovery situation: the code is
going to hit an invalid WAL record when it falls off the end of what was
correctly logged, and right now it's set up to write DEBUG messages
about any sort of invalid record. Possibly when WAL is mature we'll
consider suppressing most of the DEBUG output it currently generates...

regards, tom lane