Hi all,


I have a major problem with what seems as VSWITCH performance issue.

I have a VLAN aware VSWITCH connected to 2 X 1Gb OSA ports (1 for backup -
no link aggregation).

A continues ping between 2 Linux guest on the same VSWITCH on the same VLAN
will show high response times (meaning that all communication between the 2
guests remain within the VSWITCH)

Things like 200ms, 50ms, 8ms (not all the time but every 3 or 4 pings I get
a high one).

A continues ping from the Linux guest to itself will show much less than 1ms
all the time. I conclude from that, that the Linux guest itself is not the
problem.

It means that the Linux guest is dispatched, does not hang. It test both the
Linux responsiveness and the Linux tcpip stack.

That leads to my conclusion that the only difference of the 2 tests is the
VSWITCH being involved.

z/VM is far from 100% CPU and almost no paging is in place (10-15 pages per
second).

I can't figure out what might be the problem.

I did find 2 APARS talking about VSWITCH performance after a failover to the
backup device.

The APAR states to disconnect and connect the OSA devices (I did SET VSWITCH
DISCONNECT followed by a SET VSWITCH CONNECT) but the problem remains.



Linux guest are using Access NICs (that is not truck).



Any ideas?



Thanks

Offer Baruch


----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/

Reply via email to