On 06/11/2010 07:46 AM, Dejan Muhamedagic wrote: > Hi, > > On Fri, Jun 11, 2010 at 10:57:05AM +0200, RaSca wrote: >> Il giorno Gio 03 Giu 2010 12:29:18 CET, RaSca ha scritto: >>> Il giorno Gio 03 Giu 2010 11:48:21 CET, Lars Ellenberg ha scritto: >>> [...] >>>> I think the difference is that your client workload includes locking, >>>> and theirs do not. >>>> Please try with "nolock" on the client side, >>>> and see if that changes something. >>> There's no difference. As you can see from the hb_report, even if I'm >>> mounting with nolock nothing changes and the error is the same. Note >>> that the client from where I'm mounting is a Debian Squeeze. >> >> Finally, after a very long time, here is the solution (thanks to >> Neil Brown from NFS Mailing list). Instead of declaring a group like >> in this line: >> >> group store store-ip store-LVM store-fs store-exportfs >> >> I had to change the order: >> >> group store store-LVM store-fs store-exportfs store-ip >> >> Putting the ip at the end of the startup process and so at the begin >> of the stop process everything works and the failover is smooth even >> if I'm writing on the nfs share (the write stops for a moment and >> then restarts). >> >> It remains a little bit strange that exportfs -u does not unlock the >> exported filesystem, but removing the ip first makes things goes >> smooth, so > > Glad that it works for you now. > > Reading the Florian's configuration, just like you he also > started IP before exportfs and didn't run into any problems. What > was Neil's reasoning? > > Thanks, > > Dejan > >> Many thanks to all the guys that helped me!
Wow. Great find. I have been mucking with debian lenny init scripts, configuration files, and anything else I could think of trying to get this to work properly for well over a month now. The only difference is I am using an active/passive setup. I've simulated about 1000 manual tests, and all worked fine. Not to beat a dead cat but.... Does the removal of the IP free up (or break) the file system locks that cause the file system unmount to hang? Have we encountered a bug in the Debian implementation of NFS? (because exportfs -u doesn't work as advertised) Thanks again. Terry _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems