Re: [lustre-discuss] Lustre traffic slow on OPA fabric network
Hi Kurt, On Thu, Jul 12, 2018 at 02:36:49PM -0400, Kurt Strosahl wrote: > That's really helpful. The version on the servers is IEEL 2.5.42, while > the routers and OPA nodes are all running 2.10.4... We'be been looking at > upgrading our old system to 2.10 or 2.11. just an update on this. we moved our old 2.5 IEEL lustre to 2.10.4 (still rhel6.x) but sadly it didn't solve our lnet routing problem. sorry for the bad advice. > I checked the opa clients and the lnet routers, they all use the same > parameters that you do except for the map_on_demand (which our system > defaults to 256). we eventually realised that with the "new" ways of setting ko2iblnd and lnet options we could configure each card (qib/mlnx, opa) separately and have them "optimal", but still doesn't work without errors so far. haven't 100% ruled out shonky FINSTAR opa optical cables yet, but it seems quite unlikely. did you make any progress? cheers, robin > >w/r, >Kurt > >- Original Message - >From: "Robin Humble" >To: "Kurt Strosahl" >Cc: lustre-discuss@lists.lustre.org >Sent: Tuesday, July 10, 2018 5:03:30 AM >Subject: Re: [lustre-discuss] Lustre traffic slow on OPA fabric network > >Hi Kurt, > >On Tue, Jul 03, 2018 at 02:59:22PM -0400, Kurt Strosahl wrote: >> I've been seeing a great deal of slowness from clients on an OPA network >> accessing lustre through lnet routers. The nodes take very long to complete >> things like lfs df, and show lots of dropped / reestablished connections. >> The OSS systems show this as well, and occasionally will report that all >> routes are down to a host on the omnipath fabric. They also show large >> numbers of bulk callback errors. The lnet router show large numbers of >> PUT_NACK messages, as well as Abort reconnection messages for nodes on the >> OPA fabric. > >I don't suppose you're talking to a super-old Lustre version via the >lnet routers? > >we see excellent performance OPA to IB via lnet routers wth 2.10.x >clients and 2.9 servers, but when we try to talk to a IEEL 2.5.41 >servers then we see pretty much exactly the symptoms you describe. > >strangely direct mounts of old lustre on new clients on IB work ok, but >not via lnet routers to OPA. old lustre to new clients on tcp networks >are ok. lnet self tests OPA to IB also work fine, it's just when we do >the actual mounts... >anyway, we are going to try and resolve the problem by updating the >IEEL to 2.9 or 2.10 > >hmm, now that I think of it, we did have to tweak the ko2iblnd options >a lot on the lnet router to get it this stable. I forget the symptoms >we were seeing though, sorry. >we found the minimum common denominator settings between the IB network >and the OPA, and tuned ko2iblnd on the lnet routers down to that. if it >finds one OPA card then Lustre imposes an agressive OPA config on all >IB networks which made our mlx4 cards on a ipath/qib fabric unhappy. > >FWIW, for our hardware combo, ko2iblnd options are > > options ko2iblnd-opa peer_credits=8 peer_credits_hiw=0 credits=256 > concurrent_sends=0 ntx=512 map_on_demand=0 fmr_pool_size=512 > fmr_flush_trigger=384 fmr_cache=1 conns_per_peer=1 > >I don't know what most of these do, so please take with a grain of salt. > >cheers, >robin ___ lustre-discuss mailing list lustre-discuss@lists.lustre.org http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
Re: [lustre-discuss] Lustre traffic slow on OPA fabric network
Thanks, That's really helpful. The version on the servers is IEEL 2.5.42, while the routers and OPA nodes are all running 2.10.4... We'be been looking at upgrading our old system to 2.10 or 2.11. I checked the opa clients and the lnet routers, they all use the same parameters that you do except for the map_on_demand (which our system defaults to 256). w/r, Kurt - Original Message - From: "Robin Humble" To: "Kurt Strosahl" Cc: lustre-discuss@lists.lustre.org Sent: Tuesday, July 10, 2018 5:03:30 AM Subject: Re: [lustre-discuss] Lustre traffic slow on OPA fabric network Hi Kurt, On Tue, Jul 03, 2018 at 02:59:22PM -0400, Kurt Strosahl wrote: > I've been seeing a great deal of slowness from clients on an OPA network > accessing lustre through lnet routers. The nodes take very long to complete > things like lfs df, and show lots of dropped / reestablished connections. > The OSS systems show this as well, and occasionally will report that all > routes are down to a host on the omnipath fabric. They also show large > numbers of bulk callback errors. The lnet router show large numbers of > PUT_NACK messages, as well as Abort reconnection messages for nodes on the > OPA fabric. I don't suppose you're talking to a super-old Lustre version via the lnet routers? we see excellent performance OPA to IB via lnet routers wth 2.10.x clients and 2.9 servers, but when we try to talk to a IEEL 2.5.41 servers then we see pretty much exactly the symptoms you describe. strangely direct mounts of old lustre on new clients on IB work ok, but not via lnet routers to OPA. old lustre to new clients on tcp networks are ok. lnet self tests OPA to IB also work fine, it's just when we do the actual mounts... anyway, we are going to try and resolve the problem by updating the IEEL to 2.9 or 2.10 hmm, now that I think of it, we did have to tweak the ko2iblnd options a lot on the lnet router to get it this stable. I forget the symptoms we were seeing though, sorry. we found the minimum common denominator settings between the IB network and the OPA, and tuned ko2iblnd on the lnet routers down to that. if it finds one OPA card then Lustre imposes an agressive OPA config on all IB networks which made our mlx4 cards on a ipath/qib fabric unhappy. FWIW, for our hardware combo, ko2iblnd options are options ko2iblnd-opa peer_credits=8 peer_credits_hiw=0 credits=256 concurrent_sends=0 ntx=512 map_on_demand=0 fmr_pool_size=512 fmr_flush_trigger=384 fmr_cache=1 conns_per_peer=1 I don't know what most of these do, so please take with a grain of salt. cheers, robin ___ lustre-discuss mailing list lustre-discuss@lists.lustre.org http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
Re: [lustre-discuss] Lustre traffic slow on OPA fabric network
Hi Kurt, On Tue, Jul 03, 2018 at 02:59:22PM -0400, Kurt Strosahl wrote: > I've been seeing a great deal of slowness from clients on an OPA network > accessing lustre through lnet routers. The nodes take very long to complete > things like lfs df, and show lots of dropped / reestablished connections. > The OSS systems show this as well, and occasionally will report that all > routes are down to a host on the omnipath fabric. They also show large > numbers of bulk callback errors. The lnet router show large numbers of > PUT_NACK messages, as well as Abort reconnection messages for nodes on the > OPA fabric. I don't suppose you're talking to a super-old Lustre version via the lnet routers? we see excellent performance OPA to IB via lnet routers wth 2.10.x clients and 2.9 servers, but when we try to talk to a IEEL 2.5.41 servers then we see pretty much exactly the symptoms you describe. strangely direct mounts of old lustre on new clients on IB work ok, but not via lnet routers to OPA. old lustre to new clients on tcp networks are ok. lnet self tests OPA to IB also work fine, it's just when we do the actual mounts... anyway, we are going to try and resolve the problem by updating the IEEL to 2.9 or 2.10 hmm, now that I think of it, we did have to tweak the ko2iblnd options a lot on the lnet router to get it this stable. I forget the symptoms we were seeing though, sorry. we found the minimum common denominator settings between the IB network and the OPA, and tuned ko2iblnd on the lnet routers down to that. if it finds one OPA card then Lustre imposes an agressive OPA config on all IB networks which made our mlx4 cards on a ipath/qib fabric unhappy. FWIW, for our hardware combo, ko2iblnd options are options ko2iblnd-opa peer_credits=8 peer_credits_hiw=0 credits=256 concurrent_sends=0 ntx=512 map_on_demand=0 fmr_pool_size=512 fmr_flush_trigger=384 fmr_cache=1 conns_per_peer=1 I don't know what most of these do, so please take with a grain of salt. cheers, robin ___ lustre-discuss mailing list lustre-discuss@lists.lustre.org http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
Re: [lustre-discuss] Lustre traffic slow on OPA fabric network
It sounds like you've diagnosed the problem to be your OPA fabric. Do you have network errors that will help confirm your theory? Can you test your network without Lustre & LNet to prove its fitness? That is, do you pass network diagnostics? If it goes well, maybe LNet Self Test can help as a diagnostic. There is a guide at http://wiki.lustre.org/LNET_Selftest. -Cory -- On 7/3/18, 1:59 PM, "lustre-discuss on behalf of Kurt Strosahl" wrote: Good Afternoon, I've been seeing a great deal of slowness from clients on an OPA network accessing lustre through lnet routers. The nodes take very long to complete things like lfs df, and show lots of dropped / reestablished connections. The OSS systems show this as well, and occasionally will report that all routes are down to a host on the omnipath fabric. They also show large numbers of bulk callback errors. The lnet router show large numbers of PUT_NACK messages, as well as Abort reconnection messages for nodes on the OPA fabric. w/r, Kurt J. Strosahl System Administrator: Lustre, HPC Scientific Computing Group, Thomas Jefferson National Accelerator Facility ___ lustre-discuss mailing list lustre-discuss@lists.lustre.org http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org ___ lustre-discuss mailing list lustre-discuss@lists.lustre.org http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
[lustre-discuss] Lustre traffic slow on OPA fabric network
Good Afternoon, I've been seeing a great deal of slowness from clients on an OPA network accessing lustre through lnet routers. The nodes take very long to complete things like lfs df, and show lots of dropped / reestablished connections. The OSS systems show this as well, and occasionally will report that all routes are down to a host on the omnipath fabric. They also show large numbers of bulk callback errors. The lnet router show large numbers of PUT_NACK messages, as well as Abort reconnection messages for nodes on the OPA fabric. w/r, Kurt J. Strosahl System Administrator: Lustre, HPC Scientific Computing Group, Thomas Jefferson National Accelerator Facility ___ lustre-discuss mailing list lustre-discuss@lists.lustre.org http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org