-----Original Message----- From: Sergei Golubchik [mailto:serg@askmonty.org] Sent: Donnerstag, 4. April 2013 17:32 To: maria-developers@lists.launchpad.net Cc: wlad@mariadb.org Subject: Rev 3712: MDEV-4338 : Support atomic option on directFS/FusionIO
Hi Serg,
Hi, Vladislav!
+ srv_use_atomic_writes = (ibool) innobase_use_atomic_writes;
Is that the typical usage pattern in the innodb code? Because it looks a bit silly to me, why couldn't you create a sysvar directly on srv_use_atomic_writes, why an intermediate innobase_use_atomic_writes variable?
Yes, this is typical usage pattern in Innodb. Bit weird, I agree. Every innobase_xxx variable has an srv_xxx counterpart. I have no clue why it is so, but I did not want to disturb their existing conventions:) <skip>
it'd be nice if you could try os_file_set_atomic_writes() here to see if that works. This function creates and opens quite a few files, you could use one of them to mark it for atomic writes, and if that would fail, you'd disable atomic writes, and wouldn't change innobase_file_flush_method and innobase_use_doublewrite.
It is tricky to do it before files are opened the first time , without moving lot of code around - there is non-trivial parsing of tablespace names later on in open_or_create_data_files(), just to figure out directories and filenames. On the other hand, I think it is acceptable to fail fast during start, as long as option is not default, and good diagnostics are written into error log . look into error log, reset parameter to default. Currently there is a single filesystem with atomic capability, so people who use innodb_use_atomic_writes will probably know what they are doing. If/once the option becomes popular so it makes sense to make it default, then I would agree with you, a detection/fallback is very appropriate.
Btw, why not to use posix_fallocate whenever it's available? Or, at least, with its own --innodb-use-fallocate option?
Yes, I guess it (the new option) is a good idea. I created a followup patch that introduces the option http://lists.askmonty.org/pipermail/commits/2013-April/004569.html . I set default to ON. What do you think? Thanks a lot for looking at the patch, Wlad