Odd uuid-ossp behavior on smew and shearwater

Started by Tom Laneover 11 years ago7 messages
#1Tom Lane
tgl@sss.pgh.pa.us

Buildfarm critters smew and shearwater are reporting regression test
failures that suggest that the UUID library can't get a system MAC
address:
http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=smew&dt=2014-05-28%2023%3A38%3A28
http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=shearwater&dt=2014-05-29%2000%3A24%3A32

I've just committed regression test adjustments to prevent that from
being a failure case, but I am confused about why it's happening.
I wouldn't be surprised at not getting a MAC address on a machine that
lacks any internet connection, but that surely can't describe the
buildfarm environment. Are you curious enough to poke into it and
see what's going on? It might be useful to strace a backend that's
trying to execute uuid_generate_v1() and see what the kernel interaction
looks like exactly.

regards, tom lane

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

#2Josh Kupershmidt
schmiddy@gmail.com
In reply to: Tom Lane (#1)
1 attachment(s)
Re: Odd uuid-ossp behavior on smew and shearwater

On Wed, May 28, 2014 at 11:23 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

Buildfarm critters smew and shearwater are reporting regression test
failures that suggest that the UUID library can't get a system MAC
address:
http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=smew&amp;dt=2014-05-28%2023%3A38%3A28
http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=shearwater&amp;dt=2014-05-29%2000%3A24%3A32

I've just committed regression test adjustments to prevent that from
being a failure case, but I am confused about why it's happening.
I wouldn't be surprised at not getting a MAC address on a machine that
lacks any internet connection, but that surely can't describe the
buildfarm environment. Are you curious enough to poke into it and
see what's going on? It might be useful to strace a backend that's
trying to execute uuid_generate_v1() and see what the kernel interaction
looks like exactly.

Here's the result of attaching strace to an idle backend, then running
SELECT uuid_generate_v1(). AFAIR shearwater is a cheaply-hosted OpenVZ
VPS under the hood.

Josh

Attachments:

strace_uuid_generate_v1.txttext/plain; charset=US-ASCII; name=strace_uuid_generate_v1.txtDownload
#3Tom Lane
tgl@sss.pgh.pa.us
In reply to: Josh Kupershmidt (#2)
Re: Odd uuid-ossp behavior on smew and shearwater

Josh Kupershmidt <schmiddy@gmail.com> writes:

On Wed, May 28, 2014 at 11:23 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

I've just committed regression test adjustments to prevent that from
being a failure case, but I am confused about why it's happening.
I wouldn't be surprised at not getting a MAC address on a machine that
lacks any internet connection, but that surely can't describe the
buildfarm environment. Are you curious enough to poke into it and
see what's going on? It might be useful to strace a backend that's
trying to execute uuid_generate_v1() and see what the kernel interaction
looks like exactly.

Here's the result of attaching strace to an idle backend, then running
SELECT uuid_generate_v1(). AFAIR shearwater is a cheaply-hosted OpenVZ
VPS under the hood.

Interesting. Looks like you have access only to virtual network
interfaces, and they report all-zero MAC addresses, which the UUID library
is smart enough to ignore. If smew is also in a virtual environment
then that's probably the explanation. (There are some other buildfarm
critters that are reporting MAC addresses with the local-admin bit set,
which I suspect also means they've got virtual network interfaces, but
with a different treatment of the what-to-report problem.)

regards, tom lane

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

#4Andrew Dunstan
andrew@dunslane.net
In reply to: Tom Lane (#3)
Re: Odd uuid-ossp behavior on smew and shearwater

On 05/29/2014 02:38 PM, Tom Lane wrote:

Josh Kupershmidt <schmiddy@gmail.com> writes:

On Wed, May 28, 2014 at 11:23 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

I've just committed regression test adjustments to prevent that from
being a failure case, but I am confused about why it's happening.
I wouldn't be surprised at not getting a MAC address on a machine that
lacks any internet connection, but that surely can't describe the
buildfarm environment. Are you curious enough to poke into it and
see what's going on? It might be useful to strace a backend that's
trying to execute uuid_generate_v1() and see what the kernel interaction
looks like exactly.

Here's the result of attaching strace to an idle backend, then running
SELECT uuid_generate_v1(). AFAIR shearwater is a cheaply-hosted OpenVZ
VPS under the hood.

Interesting. Looks like you have access only to virtual network
interfaces, and they report all-zero MAC addresses, which the UUID library
is smart enough to ignore. If smew is also in a virtual environment
then that's probably the explanation. (There are some other buildfarm
critters that are reporting MAC addresses with the local-admin bit set,
which I suspect also means they've got virtual network interfaces, but
with a different treatment of the what-to-report problem.)

Almost all my critters run in VMs (all but jacana and bowerbird).

cheers

andrew

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

#5Josh Kupershmidt
schmiddy@gmail.com
In reply to: Andrew Dunstan (#4)
Re: Odd uuid-ossp behavior on smew and shearwater

On Thu, May 29, 2014 at 4:06 PM, Andrew Dunstan <andrew@dunslane.net> wrote:

On 05/29/2014 02:38 PM, Tom Lane wrote:

Josh Kupershmidt <schmiddy@gmail.com> writes:
Interesting. Looks like you have access only to virtual network
interfaces, and they report all-zero MAC addresses, which the UUID library
is smart enough to ignore. If smew is also in a virtual environment
then that's probably the explanation. (There are some other buildfarm
critters that are reporting MAC addresses with the local-admin bit set,
which I suspect also means they've got virtual network interfaces, but
with a different treatment of the what-to-report problem.)

Almost all my critters run in VMs (all but jacana and bowerbird).

They're not running on OpenVZ, are they? `ifconfig` on shearwater says:

venet0 Link encap:UNSPEC HWaddr
00-00-00-00-00-00-00-00-00-00-00-00-00-00-00-00
inet addr:127.0.0.2 P-t-P:127.0.0.2 Bcast:0.0.0.0
Mask:255.255.255.255
UP BROADCAST POINTOPOINT RUNNING NOARP MTU:1500 Metric:1
RX packets:1409294 errors:0 dropped:0 overruns:0 frame:0
TX packets:1488401 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:751149524 (716.3 MiB) TX bytes:740573200 (706.2 MiB)

venet0:0 Link encap:UNSPEC HWaddr
00-00-00-00-00-00-00-00-00-00-00-00-00-00-00-00
inet addr:198.204.250.34 P-t-P:198.204.250.34
Bcast:0.0.0.0 Mask:255.255.255.255
UP BROADCAST POINTOPOINT RUNNING NOARP MTU:1500 Metric:1

and it seems this all-zeros MAC address is a common
(mis-?)configuration on OpenVZ:

https://github.com/robertdavidgraham/masscan/issues/43
http://stackoverflow.com/questions/5838225/how-do-i-get-a-default-gridgain-node-in-openvz-discover-other-nodes-on-the-same
http://forum.openvz.org/index.php?t=msg&amp;goto=8117

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

#6Andrew Dunstan
andrew@dunslane.net
In reply to: Josh Kupershmidt (#5)
Re: Odd uuid-ossp behavior on smew and shearwater

On 05/29/2014 05:41 PM, Josh Kupershmidt wrote:

On Thu, May 29, 2014 at 4:06 PM, Andrew Dunstan <andrew@dunslane.net> wrote:

Almost all my critters run in VMs (all but jacana and bowerbird).

They're not running on OpenVZ, are they? `ifconfig` on shearwater says:

[...]

and it seems this all-zeros MAC address is a common
(mis-?)configuration on OpenVZ:

https://github.com/robertdavidgraham/masscan/issues/43
http://stackoverflow.com/questions/5838225/how-do-i-get-a-default-gridgain-node-in-openvz-discover-other-nodes-on-the-same
http://forum.openvz.org/index.php?t=msg&amp;goto=8117

Yeah, that's a bit ugly. Mine are on qemu, one (sitella) is on Xen.

cheers

andrew

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

#7Magnus Hagander
magnus@hagander.net
In reply to: Andrew Dunstan (#6)
Re: Odd uuid-ossp behavior on smew and shearwater

On Fri, May 30, 2014 at 12:45 AM, Andrew Dunstan <andrew@dunslane.net>
wrote:

On 05/29/2014 05:41 PM, Josh Kupershmidt wrote:

On Thu, May 29, 2014 at 4:06 PM, Andrew Dunstan <andrew@dunslane.net>
wrote:

Almost all my critters run in VMs (all but jacana and bowerbird).

They're not running on OpenVZ, are they? `ifconfig` on shearwater says:

[...]

and it seems this all-zeros MAC address is a common

(mis-?)configuration on OpenVZ:

https://github.com/robertdavidgraham/masscan/issues/43
http://stackoverflow.com/questions/5838225/how-do-i-
get-a-default-gridgain-node-in-openvz-discover-other-nodes-on-the-same
http://forum.openvz.org/index.php?t=msg&amp;goto=8117

Yeah, that's a bit ugly. Mine are on qemu, one (sitella) is on Xen.

That's likely the difference between real virtualization and
containerization. It'd be interesting to see how it behaves on FreeBSD
jails or LXC-containers.

--
Magnus Hagander
Me: http://www.hagander.net/
Work: http://www.redpill-linpro.com/