Hi
We have a system running Firebird 2.5.2 where the HDD / system
performance periodically drops to pathetic levels. I have just
replicated the issue.
I backed up and restored a 4GB database. Doing a sweep of the DB prior
to this caused the HDD performance to drop to 1-2mb/s rather than the
normal 100ish. End process FB and restart FB returns performance to
normal. Running a sweep on the clean / restored database is quick and
does not massively reduce performance.
So it looks like the issue is long running transactions causing a sweep
issue. However after restarting the web service that accesses this data
base adn running it for a couple of minutes is enough to cause the sweep
to be problematic again. The sweep was done with the web service disabled.
While running the database for about half an hour shows a gap between
oldest and next transaction of at most 300. My most recent run showed
the following and took more than 10 minutes (didn't wait for it to stop
before I send this email).
Flags 0
Checksum 12345
Generation 2525
Page size 8192
ODS version 11.2
Oldest transaction 1834
Oldest active 1835
Oldest snapshot 1835
Next transaction 1836
Bumped transaction 1
Sequence number 0
Next attachment ID 680
Implementation ID 26
Shadow count 0
Page buffers 1024
Next header page 0
Database dialect 3
Creation date Feb 18, 2013 9:41:47
Attributes force write
Variable header data:
Sweep interval: 40000
Any suggestions ?
Thanks
Rob
