On Mon, May 11, 2009 23:57, Sage Weil wrote: >> how much should i expext it to be working? > > Clustered mds isn't working, and the snapshots haven't been tested with > some of the recent changes, but for basic fs usage it should be > pretty stable. There are some lingering issues with the mds restarting > that have to be sorted out, but it will mostly work. > >> the kernel client was stalled several times while >> copying directories in my test setup. > > When the kernel is doing its async data writeback, metadata is also > flushed back to the mds, which makes the progress of other mds operations > stall.
hi sage! wanted to let you know about some progress i made: i changed the virtualbox network settings to use a host-internal tun/tap interface that the vboxes bridge to now. this seems better performance-wise and i did not get stalled shells or similar up to now. there is one issue though: i wrote a small zope test application (zope is a python appserver) that shows a) the last 20 db transactions and b) the last 20 created files (on a ceph mounted dir). "newtxn" creates a db txn and a file on each call. db transactions work fine, they're handled by a mysql backend. but there's an issue with files: initially there were lots of files in the folder. the zope app showed latest 20 modified of them. then i purged the whole dir on a second server with the same mount. from that time on the zope app showed still the contents of the dir before the purge (20 files). if i commit a new txn and write a new file to the folder, the reloaded page shows only the newly created files (correct dir state). if i only reload for a read, it shows old contents again. so to summarize: directory contents are out of sync accross multiple mounts/nodes. it's also in the shell, so it's not a problem of the zope app for sure! if such things happen, should there be an hints in the logfiles? can i do anything to find out what's going on here? btw one log entry that looks just not right: 09.05.12 00:44:16.207487 1150765392 -- 10.0.85.102:6800/4690/0 >> 10.0.85.102:6801/4722/0 pipe(0x1291280 sd=11 pgs=0 cs=0).connect claims to be 0.0.0.0:6801/4722/0 not 10.0.85.102:6801/4722/0 - presumably this is the same node! is that of concern? mon0/ofs0/mds ARE on the same node. just like mon1/ofs1/second mds - problem? still trying to get qemu with kvm support and bridged ethernet devices working, to have a comparison to that damn virtualbox setup. will need some further reading i guess... bes regards! jürgen -- >> XLhost.de - eXperts in Linux hosting ® << XLhost.de GmbH Jürgen Herrmann, Geschäftsführer Boelckestrasse 21, 93051 Regensburg, Germany Geschäftsführer: Volker Geith, Jürgen Herrmann Registriert unter: HRB9918 Umsatzsteuer-Identifikationsnummer: DE245931218 Fon: +49 (0)700 XLHOSTDE [0700 95467833] Fax: +49 (0)700 XLHOSTDE [0700 95467833] WEB: http://www.XLhost.de IRC: #xlh...@irc.quakenet.org ------------------------------------------------------------------------------ The NEW KODAK i700 Series Scanners deliver under ANY circumstances! Your production scanning environment may not be a perfect world - but thanks to Kodak, there's a perfect scanner to get the job done! With the NEW KODAK i700 Series Scanner you'll get full speed at 300 dpi even with all image processing features enabled. http://p.sf.net/sfu/kodak-com _______________________________________________ Ceph-devel mailing list Ceph-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ceph-devel
