my apologies if this was asked and answered before, but I just tried to get a 
HOWTO on iSCSI and multipath, but failed to find something really usable.

I have a question regarding various timeouts, specifically (SCSI command, TCP 
connection, multipath). Somehow those timeout s are hierarchical, meaning the 
multipath daemon (or kernel module) will send some command to a virtual SCSI 
device (iSCSI here), then the iSCSI layer will send a TCP packet on the wire, 
after some further unpacking a more or less real SCSI device will receive a 

If I name those timeouts tm (multipath), tt (TCP), and ts (SCSI) respectively, 
should the following be true:

tm < tt < ts

Or should the exact opposite be true:

tm > tt > ts

The problem is that the "connection" the multipath daemon sees is actually a 
multi-hop and multi-protocol connection. So multipath should probably try a 
different iSCSI/TCP connection if the network (TCP) fails to respond. 
Unfortunately iSCSI will also try some kind of connection recovery, but that 
be slower. Finally the connection could be fine, but the SCSI target could be 
unable to respond in time (e.g. persistent disk read error).

AFAIK, it's not a problem if the same iSCSI command arrives via different patch 
the iSCSI target (not saying: SCSI device), so maybe if multipath is trying a 
different patch while iSCSI is still trying TCP connection recovery could be OK.

Can anybody put some light into this (or point me to some existing docs)?


You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to open-iscsi@googlegroups.com
To unsubscribe from this group, send email to 
For more options, visit this group at http://groups.google.com/group/open-iscsi

Reply via email to