| [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] |
Well, the only reason I am concerned about lock performance is because we eval'd gfs 5.2.1, and its lock_gulm ate up system resources rather quickly due to read-calls generating lock checks. And due to the nature of the read block that occurs until the check has completed, we were getting horrible performance. Should I be concerned that opengfs (memexpd) will be plagued by these performance issues? How would opendlm handle this kinda of setup? Thanks for your advice/tips, its greatly appreciated. -Chris --- "Cahill, Ben M" <ben.m.cahill@xxxxxxxxx> wrote: > My guess would be that the locking overhead for the reads will not > impact performance very much at all. > > However, if you set it up as two separate filesystems (which I think is > what you mean by having two data partitions), you can mount one with the > "no-lock" lock protocol, and that should do what you want. As long as > you're careful not to read a given file while writing it (for example, > by disabling your service temporarily, or switching it to using the > other filesystem/partition), you should be able to successfully write to > the filesystem mounted as "no-lock". > > There are 3 methods for selecting "no-lock" for a filesystem: > > 1) filesystem creation time (writes default lock protocol into > superblock) > 2) use ogfs-tool(?) (changes default lock protocol in superblock) > 3) with mount command > > There are actually some tips for doing this in the HOWTO-opendlm doc, as > well as in ogfs man pages. > > Let us know how this works for you. > > -- Ben -- > > Opinions are mine, not Intel's > > > -----Original Message----- > > From: opengfs-users-admin@xxxxxxxxxxxxxxxxxxxxx > > [mailto:opengfs-users-admin@xxxxxxxxxxxxxxxxxxxxx] On Behalf > > Of CybeRHiDe > > Sent: Friday, February 13, 2004 5:39 PM > > To: opengfs-users@xxxxxxxxxxxxxxxxxxxxx > > Subject: [ogfs-users]Locking performance for many small reads > > > > > > My setup is a group of webservers connected via fiber channel > > to a fiber > > switch connected to a fiber channel storage array. > > > > My webservers serve many small files, with 99% of the > > activity purely being > > reads, maybe 1% writes (maintenance mainly). > > > > Could I create 2 data paritions on our storage array, one > > without locking, > > mounted ro, and one with locking, mounted rw. Then I could > > separate content > > according to the need for writes. This would substantially > > reduce the read > > lock checks that occur (correct me if I'm wrong). Is there a > > method to > > disable locking? > > > > If this scenario works, how could I make occasional updates to the ro > > partition (content updates twice a week) and not break everything? > > > > Thanks for any help you guys can offer, > > Chris > > > > > > > > ------------------------------------------------------- > > SF.Net is sponsored by: Speed Start Your Linux Apps Now. > > Build and deploy apps & Web services for Linux with > > a free DVD software kit from IBM. Click Now! > > http://ads.osdn.com/?ad_id56&alloc_id438&op=ick > > _______________________________________________ > > Opengfs-users mailing list > > Opengfs-users@xxxxxxxxxxxxxxxxxxxxx > > https://lists.sourceforge.net/lists/listinfo/opengfs-users > > > > > ------------------------------------------------------- > SF.Net is sponsored by: Speed Start Your Linux Apps Now. > Build and deploy apps & Web services for Linux with > a free DVD software kit from IBM. Click Now! > http://ads.osdn.com/?ad_id56&alloc_id438&op=click > _______________________________________________ > Opengfs-users mailing list > Opengfs-users@xxxxxxxxxxxxxxxxxxxxx > https://lists.sourceforge.net/lists/listinfo/opengfs-users __________________________________ Do you Yahoo!? Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ Opengfs-users mailing list Opengfs-users@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/opengfs-users