Re: [openib-general] .openfabrics.org names
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Jeff Squyres > Sent: Monday, January 22, 2007 4:11 PM > To: Michael S. Tsirkin > Cc: openib; Michael Paichi Lee > Subject: Re: [openib-general] .openfabrics.org names > > On Jan 22, 2007, at 4:54 PM, Michael S. Tsirkin wrote: > > >>> BTW, SSL certificate was purchased only for > staging,openfabrics.org. > >>> And, that certificate has expired. > >> > >> FWIW, I think it was just a self-signed cert. It wasn't actually > >> purchased. > > > > Maybe someone can do this for bugs.openfabrics.org for now? > > Are you asking for a self-signed cert on bugs.openfabrics.org? > > Sure, that should be do-able. Michael -- could you do that? Might I recommend a cert signed by CACert (http://www.cacert.org/)? It's no more expensive than self signed and easier to trust. -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] Mellanox SRP target implementation
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Vu Pham > Sent: Thursday, November 02, 2006 2:29 AM > To: Tomoaki Sato > Cc: openib-general@openib.org > Subject: Re: [openib-general] Mellanox SRP target implementation > > Tomoaki, > > > > > Can anybody tell me about the mellanox "SRP target" > implementation code which is included in MTD2000 with > NFS-RDMA server ? > > Is this gen2 base ? > > > > *srp target* is still on gen1 code base - IBGD > > *nfs-rdma server* is on gen2 code base Any chance the MTD2000 runs openfiler? -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] Infiniband Crossover Cable
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Adit Ranadive > Sent: Friday, October 06, 2006 10:29 AM > To: openib-general@openib.org > Subject: [openib-general] Infiniband Crossover Cable > > Im doing project in Xen+IB and wanted to connect two nodes using the > IB interconnect.. > I wanted to know if there is any kind of crossover cable available > which allows me to connect just these 2 nodes without the use of a > switch? The standard (straight-through) cables will work fine between two HCAs, just as between an HCA and a switch. Make sure at least one is hosting an SM though. -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] IPOIB failover ?
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Richard Frank > Sent: Wednesday, September 13, 2006 7:12 AM > To: Or Gerlitz > Cc: openib-general@openib.org > Subject: Re: [openib-general] IPOIB failover ? > > Supporting IPOIB fail over with the Bonding driver will work - we > currently use this for GE, etc. You can also get failover with IPoIB if you're willing to use SCTP as the transport. -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
[openib-general] [PATCH] leak in *_pingpong.c?
Be gentle, it's my first patch submission. :) The following is untested, but it looks like it's probably pretty trivial. Index: examples/rc_pingpong.c === --- examples/rc_pingpong.c (revision 9442) +++ examples/rc_pingpong.c (working copy) @@ -143,6 +143,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(servername, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for %s:%d\n", gai_strerror(n), servername, port); @@ -209,6 +210,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(NULL, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for port %d\n", gai_strerror(n), port); Index: examples/srq_pingpong.c === --- examples/srq_pingpong.c (revision 9442) +++ examples/srq_pingpong.c (working copy) @@ -154,6 +154,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(servername, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for %s:%d\n", gai_strerror(n), servername, port); @@ -233,6 +234,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(NULL, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for port %d\n", gai_strerror(n), port); Index: examples/uc_pingpong.c === --- examples/uc_pingpong.c (revision 9442) +++ examples/uc_pingpong.c (working copy) @@ -131,6 +131,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(servername, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for %s:%d\n", gai_strerror(n), servername, port); @@ -197,6 +198,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(NULL, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for port %d\n", gai_strerror(n), port); Index: examples/ud_pingpong.c === --- examples/ud_pingpong.c (revision 9442) +++ examples/ud_pingpong.c (working copy) @@ -132,6 +132,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(servername, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for %s:%d\n", gai_strerror(n), servername, port); @@ -198,6 +199,7 @@ asprintf(&service, "%d", port); n = getaddrinfo(NULL, service, &hints, &res); + free(service); if (n < 0) { fprintf(stderr, "%s for port %d\n", gai_strerror(n), port); -- -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] PXE + infiniband?
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Paul Baxter > Sent: Thursday, September 07, 2006 2:29 AM > To: openib-general@openib.org; Eli cohen > Subject: Re: [openib-general] PXE + infiniband? > > >> There is an implementation of PXE for Mellanox's HCAs that > can be found > >> here: http://sourceforge.net/forum/forum.php?forum_id=494529 > > > > Thanks for the tip > > > > I, too, am interested in this. > > > > Do you have a more direct link as I wandered around > etherboot's project > > site > > and couldn't find anything IB-specific. > > > I must have been having a 'special moment' before, because I > couldn't find > the mailing lists > > Here they are! > > http://sourceforge.net/search/?ml_name=etherboot-developers&ty > pe_of_search=mlists&group_id=4233&words=infiniband I was able to follow the procedure outlined in Eli's README and I achieved some mixed results. On one hand, lspci now shows "Expansion ROM at ed70 [disabled] [size=1M]" whereas it didn't indicate that before ("disabled" means it's zeroed out, maybe?). The BIOS seems to confirm the whole disabled thing since it doesn't list the HCA in the boot priority list. After making this change, IPoIB seems to work via this HCA, but SRP (initiation, anyways) no longer does. "ibsrpdm -c" no longer produces any output, even though I can see the target via the ibnetdiscover. Accessing the SRP target from another host on the fabric works fine. -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] PXE + infiniband?
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Cain, > Brian (GE Healthcare) > Sent: Friday, September 01, 2006 1:51 PM > To: openib-general@openib.org > Subject: [openib-general] PXE + infiniband? > > A while back > (http://openib.org/pipermail/openib-general/2005-September/010 > 801.html) > there was mention of putting PXE stuff on an HCA. Has anyone > done this > with PXELINUX? It doesn't seem like it's as straightforward as just > putting the stock PXELINUX image on your HCA. I'm assuming this image > would have to recognize the HCA and bring up IPoIB in order to use the > conventional TFTP transport? Ok, nm -- I found that etherboot has a README.boot_over_ib which looks like it'll probably work well. -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
[openib-general] PXE + infiniband?
A while back (http://openib.org/pipermail/openib-general/2005-September/010801.html) there was mention of putting PXE stuff on an HCA. Has anyone done this with PXELINUX? It doesn't seem like it's as straightforward as just putting the stock PXELINUX image on your HCA. I'm assuming this image would have to recognize the HCA and bring up IPoIB in order to use the conventional TFTP transport? -- -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] File transfer performance options
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Paul Baxter > Sent: Wednesday, August 30, 2006 2:13 PM > To: openib-general@openib.org > Cc: openib-windows@openib.org > Subject: [openib-general] File transfer performance options > > We've been testing an application that archives large > quantities of data > from a Linux system onto a Windows-based server (64bit server > 2003 R2). > > As part of the investigation into relatively modest transfer > speeds in the > win-linux configuration, we configured a Linux-Linux transfer > via IpoIB with > NFS layered on top (with ram disks to avoid physical disk issues) > > [Whilst for a real Linux-Linux configuration I would look for > the RDMA over > NFS solution, this wouldn't translate to our eventual win-linux > inter-operable system.] > > I was surprised that even on linux-linux I hit a wall of > 100MB/s (test notes > below). Are others doing better? I was hoping for 150MB/s - 200MB/s I can report streaming write results (using SRP, not NFS/IPoIB) of around 380MiB/s. Right now we think that there's a disk or controller bottleneck on the SRP target that's keeping us from getting up near 450-500 MiB/s or so. Both the initiator and target are linux-based. I think I heard of someone here using a Windows initiator and getting streaming write results similar to the 380MiB/s we're getting now. I guess there's quite a few differences in the scenarios we're describing, so it's pretty far from apples to apples. OBTW, in my experience, ext[23] seriously hamper performance. Try XFS or ReiserFS. The numbers above are all for XFS-formatted partitions. Maybe you should make the test notes a little more detailed. Doesn't NFS have a bunch of performance knobs (TCP vs UDP, block sizes, etc)? -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
[openib-general] SRP numbers from gen1 vs gen2
Does anyone have any throughput benchmark data for SRP comparing gen1 and gen2? -- -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] HCA not recognized by OFED
> -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of yipee > Sent: Thursday, August 10, 2006 7:07 AM > To: openib-general@openib.org > Subject: Re: [openib-general] HCA not recognized by OFED > > Cain, Brian (GE Healthcare ge.com> writes: > [snip] > > > I suppose I snipped a little too much when I posted the > output of lspci. > > It does look just as you indicate: "[InfiniHost III Lx HCA Flash > > Recovery]". > > What's this "Flash Recovery" in your lspci output? sounds > like you should reburn > the hca with the latest firmware. ... Yeah, apparently the flash recovery jumper was shorted against a nut on the motherboard. We clipped the leads, reflashed the card and everything works well now. Thanks for the help, everyone. -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] HCA not recognized by OFED
> -Original Message- > From: Michael S. Tsirkin [mailto:[EMAIL PROTECTED] > Sent: Tuesday, August 08, 2006 11:41 AM > To: Cain, Brian (GE Healthcare) > Cc: openib-general@openib.org > Subject: Re: HCA not recognized by OFED > > Quoting r. Cain, Brian (GE Healthcare) <[EMAIL PROTECTED]>: > > In my earlier message, I wrote the PCI ID wrong, it's not > "15b3:538d", > > it's "15b3:5e8d". I saw references to 5e8c and 5e8d sprinkled > > throughout the mthca code, but there were far more 5e8c's > than 5e8d's. > > `modinfo ib_mthca | grep -i 15b3` gives the following: > > alias: pci:v15B3d5A44sv*sd*bc*sc*i* > > alias: pci:v15B3d6278sv*sd*bc*sc*i* > > alias: pci:v15B3d6282sv*sd*bc*sc*i* > > alias: pci:v15B3d6274sv*sd*bc*sc*i* > > alias: pci:v15B3d5E8Csv*sd*bc*sc*i* > > > > ...does that mean that 5e8d is not supported? > > > > -Brian > > A modern system should show: > 5e8d MT25204 [InfiniHost III Lx HCA Flash Recovery] > > so either the flash is corrupted, or you set a jumper > to disable flash. I suppose I snipped a little too much when I posted the output of lspci. It does look just as you indicate: "[InfiniHost III Lx HCA Flash Recovery]". -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
Re: [openib-general] HCA not recognized by OFED
> -Original Message- > From: Woodruff, Robert J [mailto:[EMAIL PROTECTED] > Sent: Tuesday, August 08, 2006 11:11 AM > To: Cain, Brian (GE Healthcare); openib-general@openib.org > Subject: RE: [openib-general] HCA not recognized by OFED > > Brian wrote, > > >What other sources of info can I use to debug this problem? > > >-- > >-Brian > > Are you sure that the drivers are loaded ? > lsmod should show something like this, > > [EMAIL PROTECTED] linpack]# /sbin/lsmod | grep mthca > ib_mthca 139184 0 > ib_mad 43176 5 > ib_local_sa,ib_mthca,ib_umad,ib_sa,ib_cm > ib_core59520 14 > ib_rds,ib_srp,ib_sdp,rdma_cm,ib_local_sa,ib_ipath,ib_mthca,ib_ > ipoib,ib_u > verbs,ib_umad,ib_ucm,ib_sa,ib_cm,ib_mad Yes, they're loaded. lsmod indicates something very similar to the above. In my earlier message, I wrote the PCI ID wrong, it's not "15b3:538d", it's "15b3:5e8d". I saw references to 5e8c and 5e8d sprinkled throughout the mthca code, but there were far more 5e8c's than 5e8d's. `modinfo ib_mthca | grep -i 15b3` gives the following: alias: pci:v15B3d5A44sv*sd*bc*sc*i* alias: pci:v15B3d6278sv*sd*bc*sc*i* alias: pci:v15B3d6282sv*sd*bc*sc*i* alias: pci:v15B3d6274sv*sd*bc*sc*i* alias: pci:v15B3d5E8Csv*sd*bc*sc*i* ...does that mean that 5e8d is not supported? -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
[openib-general] HCA not recognized by OFED
I installed OFED 1.0.1 on an Intel Alcolu-based system and wasn't able to get any of the tools to recognize the HCA. It's a Mellanox (or maybe Intel?) "InfiniHost III Lx" (PSID INT_001001). lspci indicates that its PCI ID is 15b3:538d. Using Mellanox's firmware tool, I was able to detect the card, dump its firmware image and upgrade to a newer one. The card is definitely present, but when I do a ibv_devices, I get "Fatal: no infiniband class devices found." `ls /sys/class/infiniband` returns no results. Grepping for "ib_" in /var/log/* and `dmesg` returned nothing, too. Running a FC based distro, kernel 2.6.15-2.4 (SMP, x86_64). What other sources of info can I use to debug this problem? -- -Brian ___ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general