On 07/14/2010 03:59 AM, Ulrich Windl wrote:
Sean S<sstra...@gmail.com> schrieb am 13.07.2010 um 20:41 in Nachricht
<1f2389e7-9717-4f82-a05c-671f36a4c...@x21g2000yqa.googlegroups.com>:
I'm running an iscsi root partition for a CentOS machine running a
2.6.18-53 kernel. Every couple of days I get the error:
connection1:0 detected conn error (1011)
session1: session recovery timed out after 400 sec
Hi!
I cannot answer your question, but that brings up something I wanted to talk
about. Please apologize if something already exists, but I don't know:
In HP-UX 11.31 you can print "scan times" per device (i.e. LUN). Here's an
example for a true FC-SAN:
Class I H/W Path ms_scan_time
===============================
lunpath 3 0/3/1/0.0x50001fe1500c1f28.0x0 0 min 0 sec 13 ms
lunpath 24 0/3/1/0.0x50001fe1500c1f28.0x4001000000000000 0 min 0 sec 88 ms
lunpath 73 0/3/1/0.0x50001fe1500c1f28.0x4002000000000000 0 min 0 sec 88 ms
lunpath 25 0/3/1/0.0x50001fe1500c1f28.0x4003000000000000 0 min 0 sec 88 ms
lunpath 74 0/3/1/0.0x50001fe1500c1f28.0x4009000000000000 0 min 0 sec 88 ms
lunpath 26 0/3/1/0.0x50001fe1500c1f28.0x4033000000000000 0 min 0 sec 88 ms
lunpath 88 0/3/1/0.0x50001fe1500c1f28.0x4037000000000000 0 min 0 sec 88 ms
lunpath 79 0/3/1/0.0x50001fe1500c1f28.0x403d000000000000 0 min 0 sec 91 ms
lunpath 27 0/3/1/0.0x50001fe1500c1f28.0x4047000000000000 0 min 0 sec 91 ms
[...]
lunpath 63 0/7/1/0.0x500308c001d83803.0x4001000000000000 0 min 0 sec 11 ms
lunpath 64 0/7/1/0.0x500308c001d83803.0x4002000000000000 0 min 0 sec 11 ms
lunpath 65 0/7/1/0.0x500308c001d83803.0x4003000000000000 0 min 0 sec 11 ms
lunpath 66 0/7/1/0.0x500308c001d83803.0x4004000000000000 0 min 0 sec 536 ms
If Linux/open-iscsi had something similar, one could periodically watch the times to find
bottlenecks. AFAIK, the "scan time" in HP-UX is the round-trip delay for
querying a LUN or a controller (a target?).
Did you want to find bottlenecks in the network or between the initiator
and actual device or initiator and target?
Erez, was adding some code where it exports the iscsi nop/ping times.
The nop/ping we send has a header of 48 bytes and no data payload. It
does not have do any disk/device IO. So this is nice for testing the
network.
--
You received this message because you are subscribed to the Google Groups
"open-iscsi" group.
To post to this group, send email to open-is...@googlegroups.com.
To unsubscribe from this group, send email to
open-iscsi+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/open-iscsi?hl=en.