[Toolserver-l] Maintenance: user-store; all database clusters tonight

[Jan Luca]

Much better.

Thank you River!

-----Ursprüngliche Nachricht-----
Von: toolserver-l-bounces at lists.wikimedia.org
[mailto:toolserver-l-bounces at lists.wikimedia.org] Im Auftrag von River
Tarnell
Gesendet: Samstag, 11. September 2010 17:27
An: Wikimedia Toolserver Announcements
Cc: Wikimedia Toolserver Discussion
Betreff: [Toolserver-l] Maintenance: user-store; all database clusters
tonight

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

(I have the impression that our maintenance notices are not always very
clear, 
since people are sometimes surprised when we carry out previously-announced 
maintenance.  So, here's a new and hopefully improved notification format.)

This is a maintenance notification for tonight (Saturday 11th) UTC. The 
following services will be affected:

Service				| Expected impact
- --------------------------------+-----------------------------------
s1 cluster			| User databases unavailable for < 10
minutes
s2 cluster			| Entire cluster unavailable for < 10
minutes
s3 cluster			| User databases unavailable for < 10
minutes
s4 cluster			| User databases unavailable for < 10
minutes
s5 cluster			| Entire cluster unavailable for < 10
minutes
s6 cluster			| User databases unavailable for < 10
minutes
user databases (sql)		| Unavailable for < 10 minutes
user-store filesystem		| Unavailable for < 20 minutes
NFS, LDAP, misc services:	| None expected, but at risk during
maintenance.
  SVN, phpMyAdmin, SGE, tsbot,	|
  mail forwarding, DNS,		|
  sql-toolserver database	|
All other services		| No impact

Start time:	Sunday, 12th September, 12AM UTC
End time:	Sunday, 12th September, 3AM UTC

Details:

Tonight we will complete the upgrade to Solaris 10 Update 9 on the remaining

servers.  This requires under 10 minutes downtime while each system reboots.


Since we will be upgrading the primary server in each cluster, which holds
user 
databases, these will be unavailable during the reboot.  Access to
replicated
databases will not be interrupted, since queries will be directed to the 
alternative server while one is rebooted.

The exception is s2/s5 and sql (user databases), which currently have only
one 
server.  They will therefore be unavailable during the reboot.

	- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.16 (FreeBSD)

iEYEARECAAYFAkyLn7IACgkQIXd7fCuc5vLGIQCeOzuSSxfnVcwKaPz4JU4I15WL
864Ani/mUI0ed+8nfnocP3rVep+P64X6
=GEkv
-----END PGP SIGNATURE-----

_______________________________________________
Toolserver-l mailing list (Toolserver-l at lists.wikimedia.org)
https://lists.wikimedia.org/mailman/listinfo/toolserver-l
Posting guidelines for this list:
https://wiki.toolserver.org/view/Mailing_list_etiquette