On 17/04/2025 04:45, Doug Bell wrote:
There are two Bytemark servers, the first of which I don't know the
story, the second of which is donated by Bytemark care of Neil Bowers:
* cpantesters3.dh.bytemark.co.uk <http://
cpantesters3.dh.bytemark.co.uk> is 1x Intel(R) Xeon(R) CPU E3-1231 v3 @
3.40GHz / 32G memory and has about 1.5TB of hard disk space as the
previous database host. This runs the legacy bits that I haven't been
able to shut down yet. At last years summit I finally got it below a 12
load average by giving up on perfect backwards compatibility and moving
almost all of the web app to...
* cpantesters4.dh.bytemark.co.uk <http://
cpantesters4.dh.bytemark.co.uk> is 1x Intel(R) Core(TM) i3-4160 CPU @
3.60GHz / 8G memory and has about an 80G hard drive. This is serving
most of the web application these days.
Then there are two ServerCentral servers, both of which are nominally in
the hands of Joel Berger (as he still works there):
* The current primary database is 2x Intel(R) Xeon(R) CPU E5620 @
2.40GHz / 32G memory with a 2TB hard drive about 75% full of the
database data
* The backup database is 2x Intel(R) Xeon(R) CPU E5620 @ 2.40GHz / 16G
memory, and I can't find the hard drives so it's not actually backing up
anything. Will have to ask Joel to get someone to open the rack and
fiddle with it.
Finally, there's the monitoring system that I'm hosting:
* status.cpantesters.org <http://status.cpantesters.org> is a VPS with 1
core and 1 GB memory with a 40G hard drive running InfluxDB, telegraf,
and Grafana.
This is very interesting information. Thanks for providing it. Do you
think the github org could get a repository for documenting this? :)
Also, i found something interesting in the status panels which is a bit
incongruent. This page has data on jobs processed, and it also includes
a list of errors:
http://status.cpantesters.org/grafana/d/000000004/reports?orgId=1&from=1742112803397&to=1744877603397
One example of such an error is this, and the table says this job was
processed on 2025-04-16T12:46:14Z.
{"error":"'\/MooX-StrictConstructor-0.013.tar.gz' can't be parsed as a
CPAN distfile at
\/home\/cpantesters\/perl5\/lib\/perl5\/Metabase\/Resource\/cpan\/distfile.pm
line 73.\n\tMetabase::Resource::cp
What stands out here is that this distribution was uploaded on
2024-10-18, so this kind of error has likely been occurring a lot. Is
that significant and possibly related to the outages?
--
With regards,
Christian Walde
