Re: [Sks-devel] Tuning

[Jeremy T. Bouse]

On 02/11/2014 02:19 PM, Daniel Kahn Gillmor wrote:
> On 02/11/2014 01:58 PM, Benny Baumann wrote:
>> Am 11.02.2014 16:59, schrieb Kristian Fiskerstrand:
>>> Unless you run it in a clustered setup where the different members
>>> calculate it on different times and the frontend passes the request on
>>> before timeout :p
>> Its almost instantly for my maschine ...
> what is almost instantly, Benny?  Are you saying that the stats
> calculation returns almost instantly?  If so, i wonder why there is such
> a variation.  How much RAM is available for your machine?  what other
> contention do you have for disk I/O?  What kind of disks are you using?
>
> On a pretty decent machine (zimmermann.mayfirst.org), i'm seeing the
> following duration in the logs:
>
> 2014-02-11 19:17:17 Calculating DB stats
> 2014-02-11 19:17:49 Done calculating DB stats
>
> so that's over half a minute of blocked access.
>
In my case for sks.undergrid.net the result is about ~8 seconds average.
These nodes are not exactly stacked with RAM ~512MB but they're running
on SSD.

sks01
2014-02-11 17:59:59 Calculating DB stats
2014-02-11 18:00:08 Done calculating DB stats

sks02
2014-02-11 17:59:59 Calculating DB stats
2014-02-11 18:00:08 Done calculating DB stats

sks03
2014-02-11 18:00:00 Calculating DB stats
2014-02-11 18:00:06 Done calculating DB stats

In the case of keyserver.undergrid.net which has more RAM (2GB) but no
SSD the result is about the same:

2014-02-11 02:00:00 Calculating DB stats
2014-02-11 02:00:06 Done calculating DB stats

So for a 6-8 second blip I could see regenerating the stats much more
frequently. Not sure I'd do it every hour but I don't think I'd have to
wait 24 hours to do so. In the case of my sks?? nodes they only handle
SKS traffic through nginx. Besides SKS the node running keyserver runs
centralized logging with rsyslogd and a ZNC IRC bouncer daemon.

signature.asc
Description: OpenPGP digital signature