---- From: Mike Christie <michael.christie@xxxxxxxxxx> -- Sent: 2023-01-25 - 00:58 ---- > On 1/24/23 14:28, Forza wrote: >> For the past year I have been experiencing spontaneous reboots with days or weeks between each time. So eventually I got a tip to enable pstore and was able to capture several traces, and all are the same. See below. >> >> I am using Alpine Linux with kernels 5.15.x all the way up to 6.1.6, but the issue is the same. The issue seems to be mostly triggered when Windows 2016 servers that have some attached storage are rebooting during maintenance. They do not boot over iscsi, it's just extra storage for some apps on those servers. >> >> The target backingstore is using FILEIO ontop of a Btrfs filesystem (i.e. no pass-through to sas controller, etc). There is write caching enabled, but it happens even without it. I opened a report on bugzilla, earlier. There is a saveconfig.json from targetcli attached there: >> https://bugzilla.kernel.org/show_bug.cgi?id=216959 >> >> The issue seems pretty reproducible, but I am not sure how to debug further. Please advice. >> > > Are you using iscsi's multiple connections per session (MC/s) feature? No, just single sessions. > > If you just use a single connection session or multiple sessions with each > session having one connection with MPIO does it work ok? There is only one IP on clients, so I am not sure how I can configure MPIO. At least Windows does seem to allow this. https://paste.tnonline.net/files/GXulIOnf6kQV_iscsi-initiator.png I guess I have to enable more than one connection to the targets, but this seems dangerous? Thanks
