Ow Mun Heng wrote:
> I was playing around with dbi-link, hoping to get it connected to a teradata
> database. However, before I dive into that, I figured that I might as well
> try it out first on a PG Database (on another server)
>
> So, it installed dbi-link fine.
>
> I did a select on a 30GB table and it froze the Originating database and it
> ALSO froze the foreign database.
>
> Looking into the foreign box's logs, I see for some reason the network just
> ceased to function? (can't be a coincidence?)
>
> Aug 30 15:15:18 kernel: NETDEV WATCHDOG: eth0: transmit timed out
> Aug 30 15:15:18 kernel: r8169: eth0: link up
>
>
That looks like it came from dmesg. Did you look in the postgres log?
"froze" is not a helpful description. PG spawns off a client for each connection, and I doubt one client could freeze
another. So was the one connection froze, all PG clients froze, or the entire computer froze?
You said you had to reboot, so I assume the entire computer.
On the foreign box, have you ever pushed a large amount of data over the network? You might wanna try to copy some
reallybig files a few times and see if you get the eth0 timeout error again.
I assume you are using Linux and a new version of PG, right?
-Andy