Plan for seperation
by Mike McGrath
I'd like to start running our own internal DNS servers and internal load
balancers in PHX. Anyone against this? I'll be creating tickets soon.
-Mike
14 years, 7 months
Introducing Myself
by Augusto Cezar Amaral
Hi all,
My name is Augusto and I work at [1]Triforsec, An Information Security
oriented company here in [2]Recife, Pernambuco, Brazil.
I have about 7 years experience in systems administration and CCNA,
LPIC-1 and UCP-1 certifications. As a second job I teach Linux classes
at [3]Fuctura, a big training company here in Recife too.
My main insterests are Linux servers administration and IT
infrastructure/security.
I've been using Fedora/Red Hat systems since the beginning.
Cheers.
1. http://www.triforsec.com.br/
2. http://maps.google.com.br/maps?q=recife,+pernambuco,+brazil
3. http://www.fuctura.com.br/
--
Augusto Cezar Amaral da C Silva
14 years, 7 months
postgresql8.4-8.3.6
by Toshio Kuratomi
As discussed with jcollie and mmcgrath on IRC, we can take advantage of
the koji downtime to upgrade db3 (koji's db) to postgresql8.3-8.3.6.
We're currently on 8.3.3. This version should bring a many bugfixes;
none of which are affecting our ability to get things done and probably
will not result in noticable performance changes.
This release does not require a dump and reload over what we're
currrently running. That means we should be free to revert if we find
that 8.3.6 is not as reliable as 8.3.3.
This seems to be something nice to have but not essential.
I've built the packages we'll need and put them in the infrastructure
repo. jcollie, dgilmore, and others, if you want to test postgresql8.3,
feedback on whether you encounter any issues would be appreciated.
-Toshio
14 years, 7 months
New SOP page
by Toshio Kuratomi
Hey,
I've categorized all of our SOP (Standard Operating Procedure) pages in
the [[Category:Infrastructure SOPs]] and made a new page for them:
https://fedoraproject.org/wiki/Category:Infrastructure_SOPs
(The old URL redirects there so old, external links are still fine).
The index on the page is generated automatically by the category. This
means that anytime you create a new SOP simply place this Category tag
at the bottom of the page and it will appear here as well::
[[Category:Infrastructure SOPs]]
As you can see from the index on the page, the MoinMoin, hierarchical
directory index scheme makes the Category Index a little ugly
(everything is listed as beginning with "I" because they all start with
"Infrastructure". Additionally, the docs folks recommend that everyone
renames pages to be more natural language oriented so that mediawiki's
search functionality works better.
If I rename in Mediawiki, there will be redirects from the old names to
the new ones. Does anyone mind if I start doing that?
I was thinking of one of the following schemes:
A) Infrastructure/SOP/Baculs => Bacula
B) Infrastructure/SOP/Bacula => Bacula SOP
-Toshio
14 years, 7 months
Introduction
by Dmitry Kolesov
Hello.
I am sorry for my last message, coding was wrong.
And I deside to send it again.
My name is Dmitry Kolesov and I would like
to join the Infrastructure group.
I have been a Linux sysadmin for several years.
And I have experience with bash, Python, C/C++, PHP, MySQL, PostgreSQL.
I have workstations with Fedora and CentOS in my work.
I think I can help.
Best regardes.
14 years, 7 months
rsync errors on secondary
by Adrian Reber
Since 2009-02-11 I get following errors syncing from
secondary.fedoraproject.org::fedora-secondary/
rsync: send_files failed to open "/development/arm/os/Packages/.VLGothic-fonts-proportional-20080908-1.fc10.noarch.rpm.pBjp2o"
(in fedora-secondary): Permission denied (13)
rsync: send_files failed to open "/development/arm/os/Packages/.gcc-debuginfo-4.3.2-7.fa4.armv5tel.rpm.Hulh62"
(in fedora-secondary): Permission denied (13)
rsync: send_files failed to open "/development/arm/os/Packages/.gettext-devel-0.17-8.fc10.fa1.armv5tel.rpm.THlo5A"
(in fedora-secondary): Permission denied (13)
Would be great if someone could delete those temporary files.
Adrian
14 years, 7 months
keeping up with the status of alerts
by Seth Vidal
Hi folks,
I'd like to try something new. We've been getting a fair number of
alerts via nagios/zabbix/etc. Sometimes, due to the conversation in
#fedora-admin it is hard to tell who is working on what and what is in
need of immediate attention.
I'd like to suggest we use #fedora-noc for 'X is down, I'm working on
it' and discussion for immediate crisises, problems.
I sometimes have a hard time knowing what I should look at and/or what
is an expected alert due to work being done. So it would be best to keep
the noise down in #fedora-noc unless it is directly related to an
outage.
What do y'all think?
-sv
14 years, 7 months
Re: keeping up with the status of alerts
by Nigel Jones
I suggested #fedora-noc a while ago (October actually) no one seemed interested.
Ian was going to write a plugin for zodbot to echo Zabbix notices to IRC, but my understanding the code tree is dead with nothing.
I'm going to try and come up with an alternative soon.
Regards,
Nigel
----- "seth vidal" <skvidal(a)fedoraproject.org> wrote:
> Hi folks,
> I'd like to try something new. We've been getting a fair number of
> alerts via nagios/zabbix/etc. Sometimes, due to the conversation in
> #fedora-admin it is hard to tell who is working on what and what is
> in
> need of immediate attention.
>
> I'd like to suggest we use #fedora-noc for 'X is down, I'm working
> on
> it' and discussion for immediate crisises, problems.
>
> I sometimes have a hard time knowing what I should look at and/or
> what
> is an expected alert due to work being done. So it would be best to
> keep
> the noise down in #fedora-noc unless it is directly related to an
> outage.
>
> What do y'all think?
>
> -sv
>
>
> _______________________________________________
> Fedora-infrastructure-list mailing list
> Fedora-infrastructure-list(a)redhat.com
> https://www.redhat.com/mailman/listinfo/fedora-infrastructure-list
14 years, 7 months
New fasClient
by Toshio Kuratomi
For quite a while (I think sometime around the Incident) we've had a
fas-client update pending that is broken. We finally got around to
looking into the problem and have something that we think works. I've
deployed it on hosted1, hosted2, bastion, and fedorapeople. Everything
looks fine with those hosts. There are a few changes:
1) There's still one bug where old home directories aren't deleted/moved
to /home/fedora.bak/. The owners are removed, though.
2) The new script and server methods are stricter than the old one. The
new script checks for cla_done and active status whereas the old script
doesn't. This has lead to a few removals on hosted1&2, bastion, and a
whole lot of removals on fedorapeople.
For now I'm turning the fas-clients version back to "present" in puppet
and removing the known broken fas-clients packages from the fi-repo
(0.8.4.5-2 and 0.8.4.7-2, the good versions, remain). We should be good
to upgrade fas-clients on the other hosts anytime.
-Toshio
14 years, 7 months
