[Halld-offline] Fwd: New HALLDWEB1 deployment today -- will be taking everything offline shortly to get started

[Mark Ito]

Webserver switch-over today, as announced at the collaboration meeting. 
Expect outages.



-------- Forwarded Message --------
Subject: 	New HALLDWEB1 deployment today -- will be taking everything 
offline shortly to get started
Date: 	Mon, 23 Feb 2015 06:52:04 -0500 (EST)
From: 	Marty Wise <wise at jlab.org>
To: 	hovanes at jlab.org, 'Mark M. Ito' <marki at jlab.org>, davidl at jlab.org, 
'Serguei Pozdniakov' <sergpozd at jlab.org>
CC: 	'Paul Letta' <letta at jlab.org>, srwhite at jlab.org, 'Kelvin Edwards' 
<kelvin at jlab.org>



All,

Just a reminder that I will be putting the new Hall D web and DB servers 
into production today. This will require all of the involved systems to 
be taken offline until I am finished. I expect the process to take all 
day, but I will let you know when I’m done.. it’s possible that I may 
finish early. The following systems will be taken offline shortly 
(between 8:30 and 9:00) to begin the process –

HALLDWEB1

HDOPS

HALLDWEB1-2015

HALLDDB (actually an alias for HALLDWEB1 currently)

HALLDDB-2015

HALLDDB1

HALLDDB1-2015

In addition, Paul will be taking GLUONWEB offline and rebuilding it with 
RHEL7.

When things are back online, only the following systems will be available –

HALLDWEB (recall that “halldweb1” will be renamed to “halldweb”)

HALLDDB (previously an alias for halldweb1, will now be a separate system)

HALLDDB-EXT (replacing hallddb1, from off-site, it will be called 
“HALLDDB.jlab.org”)

GLUONWEB (rebuilt at RHEL7, still running the Wiki replica, etc.)

The old versions of all the systems will be retained for a couple of 
weeks for reference in case I miss anything and need them for reference. 
After things are checked out, all the old VMs will be removed.

In the central filesystems, the contents of /group/halld/www will be 
under construction as well, and should not be accessed or modified. I 
will manipulate permissions and directory names to prevent modification. 
As the deployment is completed, the directory trees will pared down, 
leaving only the new production directories. Old contents will be 
preserved, but not directly accessible to avoid confusion. The new 
directories, etc. will all match up with the corresponding machine 
names, etc.

One small point of possible confusion is with the HALLDB vs. HALLDDB-EXT 
systems. The –EXT system is the externally visible read-only replication 
slave of HALLDDB. From outside, it will be called “HALLDDB.jlab.org”. 
So, connecting to hallddb from inside will connect to the writable 
master, whereas the same connection from outside will connect to the 
read-only slave. The slave replicates everything from the master, 
including all the grants, etc. so, there should be no changes made on 
the slave… permissions changes made on the master will propagate to it.

Marty Wise (wise at jlab.org <mailto:wise at jlab.org>)
JLab IT/CNI



-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/halld-offline/attachments/20150223/b3d03576/attachment.html