capacity of keyring

[Sascha Kiefer]

Yes. It's pretty worse.

gpg: Total number processed: 569
gpg:           w/o user IDs: 3
gpg:               imported: 434  (RSA: 36)
gpg:              unchanged: 132

the program is still running:

Name                Pid Pri Thd  Hnd   Priv        CPU Time    Elapsed Time
gpg                3836   8   1   24 230048     0:14:35.265     0:15:55.496

Hmm, thats pretty bad and i have to overthink my ideas!


Holger Sesterhenn schrieb:

> Hi,
>
>> Do you know how many keys can you put into a keystore and
>> still be fast?
>> What happens when I put 10.000 keys in there? What about 100.000 keys?
>
>
> I have done some tests with dumps from a HKP keyserver (> 20MB of 
> data, 25000 keys, SuSE Linux 9.x and own Linux distribution).
> GnuPG 1.2.3 crashed during import after about 2500 keys. GnuPG 1.3.9X 
> and 1.4.0 did the job but terribly slow.
>
> It took more than 6 hours to '--import' on a P-IV/2,4 MHz (512 MB RAM, 
> 40 GB ATA)  and was only a bit faster on a dual P-IV/3 Mhz (2 GB RAM, 
> SCSI with 64 MB Cache controller).
>
> It's because GnuPG has to scan the whole keyring again and again if 
> you append keys to it. It's an exponential behaviour.
>
> GnuPG is still a client software not designed to handle such an amount 
> of keys.
>
> But as Werner mentioned, this may change in future releases ;-).