On Fri, Jan 9, 2015 at 1:31 PM,  <[email protected]> wrote:
> Hello!
>
> First of all, this is my first post to this user group.  If I'm in the wrong
> place please don't hesitate to point me in a different direction.

You got it right :)

> Starting around mid-December I've been unable to complete a backup-push.
> After running for an hour or so the server stops responding to network
> requests.  The only thing I can do is wait until backup-push finishes and
> then I can ssh back in to the server.

Maybe it's swamping everything. Try the I/O rate limiting option (see readme).

> Once back online I can find the following problems:
>
> dmesg repeats this error: [1107575.808936] xen_netfront: xennet: skb rides
> the rocket: 19 slots
> Wal-e complains about HTTP 500 when pushing files to S3 (sorry, I don't have
> a copy of this error handy)

That's potentially important. Can you make it handy?

> My server is configured as follows (let me know if more info is helpful):
>
> amazon ec2 i2.4xlarge
> ubuntu 14.04 lts
> postgres 9.3
> wal-e 7.3
> database size is ~2.4TB
>
> From what I've been able to find so far there may be a bug in the xennet
> driver that is causing the "rides the rocket" error, see here and here.
> I've tried turning some of the suggested features off with ethtool as
> suggested in the links and it seems to have prevented the "rides to the
> rocket" errors but backup-push still doesn't complete.
>
> I've since used an older backup-push to get another server going for testing
> and it too has the same problem.
>
> Has anyone else seen this?  If so, were you able to resolve it?

Nope.

Also, try the current WAL-E master. Compared to 0.7.3, I have
drastically optimized the buffer management. Performance is perhaps
even ten times better, which matters for an instance of your size.

-- 
You received this message because you are subscribed to the Google Groups 
"wal-e" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to