Jeff Breidenbach: > I applied the kernel patch "vmscan: do not unconditionally treat zones > that fail zone_reclaim() as full". Next, I activated > AUFS_CONFIG_HINOTIFY and made sure to only move data with aufs mounted > with udba=inotify. Finally, I set AuSize_DEBLK to (4 * 1024 * 1024) > and AuSize_NHASH to (16 * 1024)
Does "move data" mean "move files between branches directly" (i.e. bypassing aufs)? Otherwise you don't need to enable CONFIG_AUFS_HINOTIFY. > After this, I stopped seeing errors in dmesg. Great! However there is > still a deadlock in the filesystem. For example, an Apache process can > block for several minutes trying to read a file. During this time the > process is unkillable, load goes well over 100, aufs cannot be > remounted, and everything becomes unresponsive. This happened twice > today. I am not sure the patch for vmscan is related to the problem. If you can, try MagicSysRq + D, T, W, if you enable it. It will show which process and where in it stopped. > I tried to apply the RDU patches to > gen8.mail-archive.com:/usr/src/modules/aufs. Unfortunately the patches > did not apply cleanly, and I was afraid to continue. I am afraid your aufs1 is obsoleted. Try latest version. The RDU patch should work without the vmscan patch. J. R. Okajima ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july
