----- Original Message -----
From: "Matthew Ahrens" <[email protected]>
Thanks for taking the time to look at this again Matt much appeciated :)
These are tunables in FreeBSD and Linux does illumos have such things?
If cc optimises them away is that an issue?
mc_load: I assume this is here for debugging only? could you add a
comment to that effect?`
If you mean move this from struct to local then yes I believe this is
something that can now be done now we have the mm_preferred changes.
I don't see either of these changes in your latest webrev. Let me know
when you've made them.
Well spotted missed mc_load change, done now and will be in the next
webrev.
When you say both which other change where you refering to here?
That's not actually what it does. rotating_seek_offset is the maximum
offset at which the reduced rotating media offset will be applied,
which is currently 1/2 of rotating_seek_inc, see line:170.
Then add a comment explaining what it does control. e.g. offset of what?
reduced compared to what?
Reworded to the following, which is hopefully clear?
/*
* Offset in bytes between the current I/O and the last queued I/O which
* triggers a reduced, currently 1/2, the rotating media seek increment defined
* above.
*/
In a previous revision INT_MAX was returned if vdev_resliver_txg != 0
as it was assumed that avoiding the device that was reslivering would
improve performance. Justin found this assumption didn't hold true, hence
the early return was removed and the comment added so others are aware of
this in case they thought of the same thing.
The code should document the current behavior, not what happened in a
previous revision. It sounds like the comment should be something like
"Note that don't need to explicitly take into account the fact that this
device may be resilvering. The increased i/o load of the resilver is
already accounted for by the queue depth metric."
Sounds cool, done.
Regards
Steve
================================================
This e.mail is private and confidential between Multiplay (UK) Ltd. and the person or entity to whom it is addressed. In the event of misdirection, the recipient is prohibited from using, copying, printing or otherwise disseminating it or any information contained in it.
In the event of misdirection, illegible or incomplete transmission please
telephone +44 845 868 1337
or return the E.mail to [email protected].
_______________________________________________
developer mailing list
[email protected]
http://lists.open-zfs.org/mailman/listinfo/developer
