Quantcast
Channel: Mellanox Interconnect Community: Message List
Viewing all articles
Browse latest Browse all 6211

Re: Frequent TCP retransmissions with ConnectX-3 EN

$
0
0

Some more info.  I changed:

  options mlx4_core log_num_mgm_entry_size=-1

to:

  options mlx4_core log_num_mgm_entry_size=-7

 

in /etc/modprobe.d/mlnx.conf and restarted with "/etc/init.d/openibd restart".  Unfortunately this didn't work out too well. I am running under VMA and with this setting, VMA is crashing while connecting.  Here is the gdb backtrace:

 

(gdb) bt

#0  0x00007fa4010aa8e7 in rfs::destroy_ibv_flow() () from /usr/lib64/libvma.so

#1  0x00007fa4010a2b1d in ring::detach_flow(flow_tuple&, pkt_rcvr_sink*) () from /usr/lib64/libvma.so

#2  0x00007fa4010fe1a5 in sockinfo::detach_receiver(flow_tuple_with_local_if&) () from /usr/lib64/libvma.so

#3  0x00007fa4010fe60e in sockinfo::destructor_helper() () from /usr/lib64/libvma.so

#4  0x00007fa4011129e0 in sockinfo_tcp::connect(sockaddr const*, unsigned int) () from /usr/lib64/libvma.so

#5  0x00007fa401122b99 in connect () from /usr/lib64/libvma.so

 

-- Bud


Viewing all articles
Browse latest Browse all 6211

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>