Are the RNICs experiencing lots of pause frames during the test?
ethtool -S ethX|grep Pause
(cheiron was the server and the others were RDMA reading from it)
[r...@ajax]$ ethtool -S eth2 | grep Pause
TxPauseFrames : 248428611
RxPauseFrames : 0
[r...@achilles]$ ethtool
This email was generated automatically, please do not reply
git_url: git://git.openfabrics.org/ofed_1_4/linux-2.6.git
git_branch: ofed_kernel
Common build parameters:
Passed:
Passed on i686 with linux-2.6.16
Passed on i686 with linux-2.6.18
Passed on i686 with linux-2.6.19
Passed on i686 with
On 11:14 Fri 06 Feb , line.ho...@sun.com wrote:
I would like to get hold of the source for the 1.4 release of the
management SW.
I've tried to clone ofed_1_4/management.git, but that seems to be about 2
weeks
newer than the release.
Where / how can I find the correct version ?
Get
Hi all,
we have been struggling with the performance of a supermicro
(quad-core xeon) / qlogic (9024-FC) system running Debian, kernel
2.6.24-x86_64, and ofed-1.4 (from http://www.openfabrics.org/).
There are 8 nodes attached to the switch.
What happens is that the performance of MPI global
Sorry, the numbers for one of the tests were inserted wrongly. It should be:
***OUR-SYSTEM /supermicro-qlogic:
# OSU MPI All-to-All Personalized Exchange Latency Test v3.1.1
# SizeLatency (us)
1 137.32
2
Hi,
I would like to get hold of the source for the 1.4 release of the
management SW.
I've tried to clone ofed_1_4/management.git, but that seems to be about
2 weeks
newer than the release.
Where / how can I find the correct version ?
I was expecting to find OpenSM version 3.2.5 in the above
ucmatose allows binding to a specific address using -b. I haven't used
rds-ping
to know if it's the same as -I in that case. I don't have any systems
myself
with dual HCAs; I don't think they have enough slots to support more than
one.
Hi Sean,
ucmatose doesn't do anything with the
I played around with this a bit more yesterday - and it looks like
rdma_bind_addr()-rdma_resolve_ip()-ip_dev_find() is always returning
the first matching entry in the routing table... even though we are
providing the source ip for the bind...
Keeping in mind that both IB ports have IPs on
ucmatose doesn't do anything with the address provided with the -b param on its
--active-- side, where this problem takes place.
It passes the address into rdma_resolve_addr() as the source address, which
results in binding to that address.
- Sean
___
Hi,
I'm looking at adding pkey support into the OpenSM vendor layer. The
pkey table is a per port structure and is part of ib_port_attr_t. That
structure also include num_pkeys. There is only related API:
osm_vendor_get_all_port_attr which takes several pointers, the second
one is a pointer to a
I get these warnings trying to build with RHEL4U6 and ofa_kernel from OFED 1.4:
include/linux/jbd.h:1204:1: warning: assert_spin_locked redefined
In file included from include/linux/wait.h:25,
from include/linux/fs.h:12,
from
On Fri, Feb 6, 2009 at 2:12 PM, Hal Rosenstock hal.rosenst...@gmail.com wrote:
Hi,
I'm looking at adding pkey support into the OpenSM vendor layer. The
pkey table is a per port structure and is part of ib_port_attr_t. That
structure also include num_pkeys. There is only related API:
On Fri, Feb 6, 2009 at 8:10 PM, Sean Hefty sean.he...@intel.com wrote:
It passes the address into rdma_resolve_addr() as the source address, which
results in binding to that address.
OK, I managed to reproduce the problem with ucmatose in the same
manner it happened with rds-ping: two running
On Thu, Feb 5, 2009 at 6:47 AM, Isaac Huang he.hu...@sun.com wrote:
I got some RDMA_CM_EVENT_REJECTED errors at active sides (i.e. nodes
Poking around in CM code told me that the passive side couldn't find a
listener with
requested service_id on the incoming device of the connection request.
The variable 'offset' in iwch_sgl2pbl_map() needs to be a u64.
I assume this fixes an overflow. What's the impact of this overflow,
and when does it trigger? ie is this urgent enough for 2.6.29 maybe?
- R.
___
general mailing list
Roland Dreier wrote:
The variable 'offset' in iwch_sgl2pbl_map() needs to be a u64.
I assume this fixes an overflow. What's the impact of this overflow,
and when does it trigger? ie is this urgent enough for 2.6.29 maybe?
- R.
This was actually found by a customer using another OS
Roland Dreier wrote:
+ BUG_ON((*cqe_flushed == 0) !SW_CQE(*hw_cqe));
BUG_ON()s are kind of nasty -- possibly killing the whole box because of
a driver issue or an unanticipated HW quirk -- is there any way to
report this problem and try to limp on?
I'm not sure I agree with
I'm not sure I agree with trying to limp on. This BUG_ON() doesn't
indicate a HW quirk. It indicates the driver logic is busted. Isn't
that what BUG_ON() should be used for?
Yeah, I guess so -- the only issue is that it's very annoying for some
buggy driver to kill the whole system when
applied 1-2
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
thanks, applied
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
thanks, applied
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Roland Dreier wrote:
I'm not sure I agree with trying to limp on. This BUG_ON() doesn't
indicate a HW quirk. It indicates the driver logic is busted. Isn't
that what BUG_ON() should be used for?
Yeah, I guess so -- the only issue is that it's very annoying for some
buggy driver to kill
On Fri, Feb 06, 2009 at 12:05:48PM -0500, Richard Frank wrote:
I played around with this a bit more yesterday - and it looks like
rdma_bind_addr()-rdma_resolve_ip()-ip_dev_find() is always returning the
first matching entry in the routing table... even though we are providing
the source ip
@@ -752,6 +752,11 @@ struct ib_send_wr {
int access_flags;
u32 rkey;
} fast_reg;
+struct {
+struct ib_unpacked_lrh *lrh;
+
-type != IB_QPT_SMI type != IB_QPT_GSI)
+type != IB_QPT_SMI type != IB_QPT_GSI type != IB_QPT_RAW_ETY)
Seems we're at the point where mlx4 could use a is_special_qpt()
helper maybe?
err = create_qp_common(dev, pd, init_attr, udata,
On Fri, Feb 6, 2009 at 1:47 AM, Hal Rosenstock
h...@obsidianresearch.com wrote:
Sasha,
This patch sets the attribute ID based on what is in the response.
Hal,
Your patches can't really be reviewed when being sent as attachment,
any reason not
to send them embedded within the email message?
Interesting - Andy Grover pointed this out too - and I totally (as
usual) missed the point. :(
Jason Gunthorpe wrote:
On Fri, Feb 06, 2009 at 12:05:48PM -0500, Richard Frank wrote:
I played around with this a bit more yesterday - and it looks like
I want to install the OFED on a PC without any infiniband devices ,can this
idea work? or what i need (hardware) when i want to install this software on
a general computer ?
___
general mailing list
general@lists.openfabrics.org
You can install OFED fine on a supported operating system with no
present Infiniband device, but the tools will generally not function
without an Infiniband device.
Cameron
wrote:
I want to install the OFED on a PC
without any infiniband devices ,can this idea work? or what i need
29 matches
Mail list logo