Description of problem: DM-Multipather with initialization problem ? After setup of 40 vdisks, zoning of two switches, etc. and starting I/O I did a goodpath-test with Redhat 5.1. Everything worked fine. During running the failover test's the dm_multipather didn't handle to re-establish all paths correctly. Host: Itanium 2; IBM x455 Server. I tried running multipathd with "-v4" option for verbose logging. And changed the line "daemon $DAEMON" with "daemon $DAEMON -v4" in /etc/init.d/multipathd file and restarted the daemon by running "/etc/init.d/multipathd restart" again. After that the failover worked fine, with and without the -v4 option. In my setup this behavior is reproducible. Could there be a connection with specifically starting the multipathd ? Kind regards, Achim Version-Release number of selected component (if applicable): Linux x455-tic-1.de.ibm.com 2.6.18-53.el5 #1 SMP Wed Oct 10 16:34:55 EDT 2007 ia64 ia64 ia64 GNU/Linux multipath-tools v0.4.7 (03/12, 2006) How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Reassigning to proper component. Read ya, Phil
You need to start the daemon to have the paths checked & re-instated. The in-kernel support just handles removing paths when I/O errors occur.
This bug has been dormant in NEEDINFO for quite a long time. Looks like user error (daemon wasn't started and/or service wasn't enabled). Closing NOTABUG.