Brad Hall wrote: > On Wed, Sep 24, 2008 at 10:56:03AM -0700, [EMAIL PROTECTED] wrote: > >> On Wed, Sep 24, 2008 at 10:35:22AM -0700, Brad Hall wrote: >> >>> On Wed, Sep 24, 2008 at 10:24:29AM -0700, [EMAIL PROTECTED] wrote: >>> >>>> Hey Brad, >>>> >>>> I reviewed this before and still think it looks good. >>>> >>>> The one suggestion I had was to add EIO to the efail.c test case, since >>>> ZFS can return an EIO in situations where it determines that data has >>>> been corrupted and is unable to recover. (No replica exists) >>>> >>>> Based upon our previous conversations, you said that this error code >>>> currently confuses the test-suite. I still think it's a worthwhile case >>>> to test. Perhaps you could file a bug to enhance this case to support >>>> EIO when it's plausible? >>>> >>> Just added this and updated the webrev. I'm planning to also make another >>> test-target that does EIO via zinject but that isn't done yet, so I put EIO >>> in >>> here for now. >>> >> I'm a little confused. I thought you had said that sending an EIO from >> this code would cause bad things to happen in the test harness. Did >> that turn out not to be the case? >> > > Yeah, that's not the case now: we only [return -1, set errno] when the PKG env > variable is set (and it's only set when running pkg commands). > > Thanks, > Brad > _______________________________________________ > pkg-discuss mailing list > [email protected] > http://mail.opensolaris.org/mailman/listinfo/pkg-discuss > Looks good to me.
Nits: efail.c line 43: I think a g is missing at the end of the line Random questions: Could we, or do we even want, to be able to specify a parameter on the command to adjust the percentage of reads or writes which produce errors? Might we ever want to only test problems in reading, or writing, or making dirs separately from the other errors? (Could this be a command line argument too?) That's all I've got, Brock _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
