> I read somewhere that it does its work...slowly. I > presume the "quotas" > file is sparse, so couldn't quotacheck use the new > SEEK_DATA/SEEK_HOLE > functionality to skip past holes rather than having > to read in and skip over > so many zero records from the "quotas" file? > > Maybe I'm off my rocker, but if it could be done, it > seems like it would > be a heck of a lot faster. > > Given that it looks like quotacheck uses stdio, it > might help if fseek(3c) > could also support SEEK_DATA/SEEK_HOLE. The code > doesn't seem to me > to preclude it, but I haven't thought through whether > or not it would behave > sanely with those. And at least as of Solaris 10, > the man page didn't mention > the possibility. Also, I have no idea why > http://src.opensolaris.org/source/xref/onnv/onnv-gate/ > usr/src/lib/libc/port/stdio/fseek.c#66 > uses the constant 1 rather than SEEK_CUR.
yepp, there's a lot not state of the art with UFS quotacheck, how it works, and its particular slowness: it would really need some love since ages, not only to speed up quotacheck in particular, but also to improve the overall infrstructure it uses: 4665378 Slow quotacheck/Scanning large quotafile causes longer than necessary bootup time http://bugs.opensolaris.org/view_bug.do?bug_id=4665378 4741624 need asynchronous in-kernel replacement for quotacheck http://bugs.opensolaris.org/view_bug.do?bug_id=4741624 CC'd ufs-discuss for the records. --- frankB This message posted from opensolaris.org _______________________________________________ opensolaris-code mailing list opensolaris-code@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
