On Fri, Sep 19, 2008 at 04:57:32PM -0500, Andy wrote: > On Fri, Sep 19, 2008 at 01:53:13PM -0500, Andy wrote: > > On Fri, Sep 19, 2008 at 03:27:10AM -0000, bmarzins@xxxxxxxxxxxxxx wrote: > > > CVSROOT: /cvs/dm > > > Module name: multipath-tools > > > Branch: RHEL5_FC6 > > > Changes by: bmarzins@xxxxxxxxxxxxxx 2008-09-19 03:27:08 > > > > > > Modified files: > > > libmultipath : configure.c configure.h devmapper.c devmapper.h > > > multipathd : cli.c cli.h cli_handlers.c cli_handlers.h > > > main.c main.h > > > > > > Log message: > > > Fix for bz #455692. There is now a command for multipathd, 'resize map'. This > > > does online grows or shrinks for your multipath device. It has some warts. > > > > > > 1. It needs to remove and re-add every path in the device. This can cause two > > > annoyances. First, if you only have one valid path, you go into an all paths > > > down scenario. However, you will immediately re-add the path, so it won't > > > trigger problem #2. Second, if the path is inaccessable, you may not be able to > > > re-add it. > > > > > > 2. You can't disable flushing on suspend for a device-mapper device, when you > > > want to change size. This means that if all your paths are down, you will fail > > > the queued IOs when you suspend. This is a problem with the device-mapper kernel > > > code, and there's no real way to avoid it, except by fixing it in the kernel. > > > > > > This commit also has some minor fixups for 238421, and adds the ability to > > > disable and restore queueing on all multipath devices through multipathd > > > > > > Patches: > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/libmultipath/configure.c.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.2.2.3&r2=1.2.2.4 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/libmultipath/configure.h.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.1&r2=1.1.2.1 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/libmultipath/devmapper.c.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.22.2.5&r2=1.22.2.6 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/libmultipath/devmapper.h.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.11.2.4&r2=1.11.2.5 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/multipathd/cli.c.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.5.2.2&r2=1.5.2.3 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/multipathd/cli.h.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.5.2.2&r2=1.5.2.3 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/multipathd/cli_handlers.c.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.6.2.2&r2=1.6.2.3 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/multipathd/cli_handlers.h.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.3.2.2&r2=1.3.2.3 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/multipathd/main.c.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.69.2.13&r2=1.69.2.14 > > > http://sourceware.org/cgi-bin/cvsweb.cgi/multipath-tools/multipathd/main.h.diff?cvsroot=dm&only_with_tag=RHEL5_FC6&r1=1.7&r2=1.7.2.1 > > > > > > > Where can I get the multipath-tools sources that these patches apply cleanly > > to? They do not apply cleanly to my multipath 0.4.8 sources, add there > > seems to be some major differences in some places. > > > > Thanks, > > > > Andy > > > > Responding to my own mail, I did got the source with the patches from the > cvs. But, the resizing does not work for with with mounted volumes. There > are 4 paths for the volume, so there is no probably with it failing paths. > I get "domap(0) failure for create/reload map" when I run multipath -v 3. > > I can get dm to reconize the new size if the volume is umounted. But, I > really would like it to work while it is mounted. > It should work for mounted volumes. Did you try multipathd -k"resize map <mapname>" Since this requires suspending without the no_flush option, I didn't want to have it happen on a normal reload, and since without the kernel fixes, this required removing and then readding paths, I did it through multipathd. So just running multipath -v3 after a path resize won't work. In fact, I think it might leave the devices in the suspended state, since the resume fails. -Ben > Andy > > -- > dm-devel mailing list > dm-devel@xxxxxxxxxx > https://www.redhat.com/mailman/listinfo/dm-devel -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
