On Mon, 2014-05-12 at 15:20 -0400, deeepdish wrote: > Hello, > > We implemented LIO using Fedora 20 to host storage for our backup > environments. For this we use bcache + LVM to achieve a decent > performing write optimized storage pool. > > There are two storage nodes in our environment both of which are > identical in hardware and OS configuration, however only one is giving > us some grief when it comes to LIO. We're seeing a lot of LIO based > activity around timeouts, errors, and target disconnects. > To confirm, you've got the DS4000 branded ISP2422 HBAs in both machines configured with tcm_qla2xxx target ports, and one machine is able to reproduce the timeouts below while the second machine appears stable, correct..? Also, is it both ISP2422 HBAs that start having problems when the issue occurs, or only one of them..? > In an attempt to remedy these issues we attempted to upgrade the > kernel several times, however after some seeming success, the issue > whereby the Qlogic Target looses connection to backing LUNs for > whatever reason. During this time, the volumes in question maintain > their availablity in LVM and can be seen by the Linux host serving as > the FC target. > > This particular case occured 9 days after our last reboot / patch of > the server. This time around we upgraded the kernel to 3.14.2. I > captured the following logs / pre- / post- failure to provide as much > detail as possible. So if the target machines are identical, then it would most likely mean some type of hardware issue, given the age of the systems involved. Eg: It doesn't appear to be a software issue specific with ISP2422 HBAs. That said, I'd consider isolating individual components (HBA, PCI-X slot, GBICs, etc) to further help identify the problem space. --nab -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
