Early BSD status report

[rleigh]

Dear Jim,

Regarding LD_LIBRARY_PATH, I've also had to do this when building with GCC. The base libgcc is from 4.2 and lacks features needed by code compiled with later GCC releases such as exception handling. I saw a post on the FreeBSD lists last week about rectifying this. See https://lists.freebsd.org/pipermail/fre ... hread.html thread "Problems with our libgcc_s.so in base". My approach for our C++ code has been to support the base toolchain only (clang 3.4; will be 3.8 once 11.0-RELEASE is out), and to ensure that all libraries and programs are built with this. If you can avoid building dependent Python bits with GCC, it might help.

Regarding the local loopback, it works for me for v4 but not for v6:

Code: Select all

amys% jls
   JID  IP Address      Hostname                      Path
     1  192.168.1.13    farstrider.codelibre.net      /jail/postgres
     2  192.168.1.15    bfcpp-bare.codelibre.net      /jail/bfcpp-bare
     3  192.168.1.12    bfcpp.codelibre.net           /jail/bfcpp
     4  192.168.1.11    bair.codelibre.net            /jail/kfreebsd-unstable

bfcpp% ifconfig
bge0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
        options=c019b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4,VLAN_HWTSO,LINKSTATE>
        ether 38:ea:a7:ab:61:53
        inet 192.168.1.12 netmask 0xffffffff broadcast 192.168.1.12
        inet6 2001:8b0:860:ddbd:3aea:a7ff:feab:7002 prefixlen 128
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>
        media: Ethernet autoselect (100baseTX <full-duplex>)
        status: active
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384
        options=600003<RXCSUM,TXCSUM,RXCSUM_IPV6,TXCSUM_IPV6>
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>

bfcpp% ping -c1 127.0.0.1
PING 127.0.0.1 (127.0.0.1): 56 data bytes
64 bytes from 127.0.0.1: icmp_seq=0 ttl=64 time=0.055 ms

--- 127.0.0.1 ping statistics ---
1 packets transmitted, 1 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 0.055/0.055/0.055/0.000 ms
bfcpp% ping6 -c1 ::1     
PING6(56=40+8+8 bytes) 2001:8b0:860:ddbd:3aea:a7ff:feab:7002 --> ::1
ping6: sendmsg: Can't assign requested address
ping6: wrote ::1 16 chars, ret=-1

--- ::1 ping6 statistics ---
1 packets transmitted, 0 packets received, 100.0% packet loss


Looks like https://lists.freebsd.org/pipermail/fre ... 01500.html but I can't see anything more recent. Edit: also see https://lists.freebsd.org/pipermail/fre ... 03179.html where I've queried the status of this issue.

Regarding multicast, have you seen https://bugs.freebsd.org/bugzilla/show_ ... ?id=180067 ?

I haven't tried vnet to see if this offers any improvement in this respect. Once 11.0-RELEASE is out, I'll upgrade the above system and have a play with it.


Regards,
Roger

[jimlong]

Thanks for your reply, Roger.

My intent was to avoid building anything with GCC at all, but GCC gets pulled in as a dependency of numpy, for reasons I don't yet understand.

Regarding the loopback interface, yes, the IP is pingable from within the jail, but to my knowledge, the jail cannot bind a socket to the 127.0.0.1 IP, unless a full virtual network stack is instantiated inside the jail.

Thank you also for the link to tracker issue 180067. That looks like a show-stopper, at least for a jailed implementation of Omero in the near term. I am inclined to park this FreeBSD installation for the moment and attempt a CentOS 7/Ice 36 installation just to regain some momentum and also get a feel for how a successful installation proceeds! I have hit one bump in the CentOS installation guide, but that's a matter for a separate thread.

Jim