[OpenAFS] Re: Server disk operations speed

[Jukka Tuominen]

(list cc:ed)


Thanks for the hint Timothy,

I have actually tried this earlier, but since some things have changed
since then, I gave it a new try. I tried it within the server machine
(server & client), and with a remote client via WLAN.

It doesn't seem to be the bottleneck. The change is relatively small, and
I'm kind of glad because encryption is a must since this system is WAN
based. LAN is only development phase luxory.

br, jukka



> Jukka,
>
> Have you tried turning off encryption?
>
> fs setcrypt off
>
> This rids a tremendous amoutn of overhead, and I simply do not use it in
> my
> facility since all of our AFS servers are in a secure network.
>
> Best,
>
> --Timothy
>
>
>
> On Wed, Apr 10, 2013 at 10:51 AM, Jukka Tuominen <
> jukka.tuominen@finndesign.fi> wrote:
>
>>
>>
>> > On Tue, 9 Apr 2013 23:36:15 +0300 (EEST)
>> > "Jukka Tuominen" <jukka.tuominen@finndesign.fi> wrote:
>> >
>> >> Does it sound resonable to prefer DNS names over external IP's, and
>> >> external IP's over local IP's? The rationale there would be to prefer
>> >> dynamic over static definitions, and IP's that the clients can see
>> >> over local IP's that may not be accessible always? (external meaning
>> >> WAN and internal LAN IP's). I've started with whatever worked, and
>> >> later on moved towards what I just described.
>> >
>> > I'm not quite sure if you're asking what you as a person should do, or
>> > if you're asking what the code does.
>> >
>> > If you mean what IPs you should be specifying in various places, you
>> > should put in the 'public' IPs that are always reachable by everyone.
>> > That means put that in CellServDB and probably the fileserver NetInfo.
>> > If you put a 'private' IP in there, then clients accessing from
>> > elsewhere may also see that IP, and obviously they won't be able to
>> > connect. We don't really do "split-horizon"-style setups, since
>> everyone
>> > gets the same list of IPs for fileservers.
>>
>> This answered my question. These are my initial findings with Wireshark:
>>
>> When running wireshark on the idle server machine, there was no striking
>> errors that I could spot while browsing through the output. I then
>> grepped
>> /etc/ for any private IP entries. Since the machine is on DMZ, it has a
>> private address in /etc/network/interfaces.
>> /etc/openafs/server/CellServDB
>> and /etc/openafs/CellServDB were the only other ones. I changed the
>> latter
>> two to public ones, but wireshark started to print "unreachable ports".
>> It
>> seemed that /etc/openafs/server/CellServDB had to have the private IP in
>> place in order not to raise errors.
>>
>> It also seems, that there is quite a frequent traffic between the
>> private
>> and the public IP. I believe the traffic goes to the NIC at least, but
>> propably not further (hub, switch or firewall) but I can't say for sure.
>> Wireshark pairs the private and public, and then public and private on
>> single lines, so how is that interpreted?
>>
>> I'd like to get rid of that noise because it may be _the_ problem, or at
>> least it won't speed up things for sure. I'm just not sure whether it is
>> possible from DMZ. Does anybody else have afs server on DMZ, and still
>> get
>> the full speed. And if yes, then do you have a private or public IP
>> assigned?
>>
>> Or maybe I have just routed the traffic otherwise backwards, somehow...
>>
>>
>> br, jukka
>>
>> >
>> >> If you just want to see where the traffic is
>> >> > going, capture some net traffic. All AFS traffic for a client
>> >> > interacting with a server should happen on ports 7000-7003 UDP. If
>> >> > we're contacting an IP that is one of the local host's interfaces,
>> >> > it's up to the underlying OS where the packets physically go.
>> >>
>> >> OK, I'll try to figure out how to do that. Wireshark is my guess to
>> >> start with.
>> >
>> > Yes, tcpdump or wireshark are the usual tools. wireshark is/can be
>> > graphical and probably more intuitive if you haven't used similar
>> tools
>> > before.
>> >
>> > --
>> > Andrew Deason
>> > adeason@sinenomine.net
>> >
>> > _______________________________________________
>> > OpenAFS-info mailing list
>> > OpenAFS-info@openafs.org
>> > https://lists.openafs.org/mailman/listinfo/openafs-info
>> >
>>
>>
>> _______________________________________________
>> OpenAFS-info mailing list
>> OpenAFS-info@openafs.org
>> https://lists.openafs.org/mailman/listinfo/openafs-info
>>
>
>
>
> --
> Timothy Balcer / IT Services
> Telmate / San Francisco, CA
> Direct / (415) 300-4313
> Customer Service / (800) 205-5510
>