MirrorManager Infrastructure SOP

From FedoraProject

(Difference between revisions)
Jump to: navigation, search
(Initial MirrorManager SOP)
 
Line 8: Line 8:
 
Location: Phoenix
 
Location: Phoenix
  
Servers: app3, app4, app5, app6
+
Servers: app1, app2, app3, app4, app5, app6, bapp1
  
 
Purpose: Manage mirrors for Fedora distribution
 
Purpose: Manage mirrors for Fedora distribution
Line 20: Line 20:
 
==== Regenerating the Publiclist ====
 
==== Regenerating the Publiclist ====
  
On app4:
+
On bapp1:
  
 
<pre>
 
<pre>
Line 29: Line 29:
  
 
<pre>
 
<pre>
/srv/tg/mirrormanager/update-each-server > /tmp/mirrormanager-mirrorlist.log 2>&1 && \
+
/usr/share/mirrormanager/server/update-each-server > /tmp/mirrormanager-mirrorlist.log 2>&1 && \
/srv/tg/mirrormanager/mm_sync_out
+
/usr/share/mirrormanager/mm_sync_out
 
</pre>
 
</pre>
  
Line 39: Line 39:
 
rm -rf /srv/cache/mod_cache/*
 
rm -rf /srv/cache/mod_cache/*
 
</pre>
 
</pre>
 +
 +
==== Hung admin.fedoraproject.org/mirrormanager ====
 +
 +
This generally happens when an app server loses connection to db2.
 +
 +
# on bapp1 and app[1-6], su up, kill the start-mirrormanager.py script.  supervisor will restart it
 +
# on bapp1, if crawlers and update-master-directory-list are likewise hung, kill them too.

Revision as of 13:48, 9 October 2008

Contents

MirrorManager - SOP

Contact Information

Owner: Fedora Infrastructure Team

Contact: #fedora-admin, sysadmin-main, sysadmin-web

Location: Phoenix

Servers: app1, app2, app3, app4, app5, app6, bapp1

Purpose: Manage mirrors for Fedora distribution

Description

TODO

Troubleshooting and Resolution

Regenerating the Publiclist

On bapp1:

sudo su - mirrormanager

then

/usr/share/mirrormanager/server/update-each-server > /tmp/mirrormanager-mirrorlist.log 2>&1 && \
/usr/share/mirrormanager/mm_sync_out

To make this take effect immediately, you may need to remove the cache on the proxies:

# As root on proxy[1-3]
rm -rf /srv/cache/mod_cache/*

Hung admin.fedoraproject.org/mirrormanager

This generally happens when an app server loses connection to db2.

  1. on bapp1 and app[1-6], su up, kill the start-mirrormanager.py script. supervisor will restart it
  2. on bapp1, if crawlers and update-master-directory-list are likewise hung, kill them too.