Subject | Async writes/forced writes |
---|---|
Author | Kjell Rilbe |
Post date | 2005-03-23T13:28:59Z |
Hi,
In a comment to a feature request, Dmitry Yemanov writes this:
Win32 I should turn forced writes on to avoid db corruption due to
potentially indefinately delayed cache flushes (caused by the OS).
So, is it true? Specifically:
- Is it true that this feature exists already, in v1.5?
- Is it true that it really works on Win32?
- Any risks I should be aware of, apart from the obvious one: risk of DB
corruption on crash/powerdown during MaxUnflushedWriteTime?
Thanks,
Kjell
--
--------------------------------------
Kjell Rilbe
Adressmarknaden AM AB
E-post: kjell.rilbe@...
Telefon: 08-761 06 55
Mobil: 0733-44 24 64
In a comment to a feature request, Dmitry Yemanov writes this:
> Since v1.5, async writes are controlled by the engine. The cacheI've never heard this before. What I've *repeatedly* heard is that on
> manager flushes dirty pages accordingly to MaxUnflushedWrites and
> MaxUnflushedWriteTime configuration parameters. This gives you a
> compromise between fast-but-never-flushed (old FW=OFF) and
> low-but-reliable (FW=ON) page write mechanisms.
Win32 I should turn forced writes on to avoid db corruption due to
potentially indefinately delayed cache flushes (caused by the OS).
So, is it true? Specifically:
- Is it true that this feature exists already, in v1.5?
- Is it true that it really works on Win32?
- Any risks I should be aware of, apart from the obvious one: risk of DB
corruption on crash/powerdown during MaxUnflushedWriteTime?
Thanks,
Kjell
--
--------------------------------------
Kjell Rilbe
Adressmarknaden AM AB
E-post: kjell.rilbe@...
Telefon: 08-761 06 55
Mobil: 0733-44 24 64