fs full stops postmaster

Started by Zeugswetter Andreas SBabout 25 years ago2 messages
#1Zeugswetter Andreas SB
ZeugswetterA@wien.spardat.at

When the filesystem fills and the txlog cannot be written, then the postmaster dies.

postgres@s0188000zeu:/usr/postgres> time echo "copy journaleintrag from '/tmp/j.unl' delimiters '|';" | psql cusejoua
FATAL 2: copy: line 64902, write(logfile 0 seg 2 off 6094848) failed: No space left on device
pqReadData() -- backend closed the channel unexpectedly.
This probably means the backend terminated abnormally
before or while processing the request.
connection to server was lost

I think it shoud stay up and running, but deny any modification until the filesystem is extended.

Andreas

#2Zeugswetter Andreas SB
ZeugswetterA@wien.spardat.at
In reply to: Zeugswetter Andreas SB (#1)
AW: fs full stops postmaster

When the filesystem fills and the txlog cannot be written,
then the postmaster dies.

postgres@s0188000zeu:/usr/postgres> time echo "copy
journaleintrag from '/tmp/j.unl' delimiters '|';" | psql cusejoua
FATAL 2: copy: line 64902, write(logfile 0 seg 2 off
6094848) failed: No space left on device
pqReadData() -- backend closed the channel unexpectedly.
This probably means the backend terminated abnormally
before or while processing the request.
connection to server was lost

I think it shoud stay up and running, but deny any
modification until the filesystem is extended.

This is an effect of the fact, that the file is not really created in its full size
even if ls shows the 16 mb. As an additional effect, the db does not
start up any more --> txlog corrupted.

Andreas