Hi Mike: Thank you for your reply.
On Friday, September 21, 2018 at 9:35:25 AM UTC-7, Mike Christie wrote: > > > ... > > This is one of those things where I said I would fix if anyone ever > complains, but luckily I made it through my entire time as maintainer > and no one ever did. So, I guess basically tag you're it :) > > Checkout node.session.initial_login_retry_max and iscsi_login_eh. The > current state machine detects the iscsid restart case as a relogin and > so that setting does not kick in. You could: > > 1. modify the state machine check so it does. > 2. add a new setting that works like initial_login_retry_max but doesn't > really work like it :) What I mean is that initial_login_retry_max was > trying to handle a lot of different cases so it is a little odd. See the > iscsid.conf comments. You probably want to make something that is more > straightforward. > I didn't see this reply until this morning, but it's funny that I came to similar conclusions. Even though Chris and I "own" this code now, I don't know it well enough, so I spent a bunch of time learning what is going on here now. And the libopeniscsiusr library complicates things a bit because some functionality is now in two places! (Not sure how I let that happen, but that's a different story.) I actually picked number 2. I created session.relogin_max for the new value because there was already a similar value for discovery. So now the daemon code, when trying to recover a session, will only retry "relogin_max" times, by default 32. But this still isn't an ideal long-term solution, because this is _per-session_. It's easy to imagine hundreds of sessions. This should probably be done asynchronously, and it probably shouldn't keep the daemon from coming completely up. But the good news is that now, after about a minute, the stale session goes away if the target can't be reached. And if the target comes back during that minute, the session is correctly reestablished. I also fixed some bugs so that we can see the session correctly during this period of stale-session-connection-retry, and we will get a message along the lines of "target temporarily not connected" if we try to logout of one of these stale sessions. I will pose this solution as a pull request on github "real soon now". :) -- You received this message because you are subscribed to the Google Groups "open-iscsi" group. To unsubscribe from this group and stop receiving emails from it, send an email to open-iscsi+unsubscr...@googlegroups.com. To post to this group, send email to open-iscsi@googlegroups.com. Visit this group at https://groups.google.com/group/open-iscsi. For more options, visit https://groups.google.com/d/optout.
