From: Tristan C Collier <[log in to unmask]>
Date: Fri, 17 Feb 2012 14:11:41 +0000

Dear All,

Further to our previous message, we can now provide more detail.

The performance problems experienced on CJO this week are due to
corruption in the access control cache system. This is a critical part
of the CJO system - just about every page checks access - and is
fundamental to adequate site performance. If we have corruption the
system bypasses this cache and the load increases significantly
causing the slowdown and outages experienced by some customers.

In terms of immediate actions we either have done or are planning the following:

1.  On Wednesday 15th February, after the most significant set of
problems, we amended a number of application settings. This has
stabilised the situation allowing us to continue with further work on
the system without affecting users.

2.  We will need to rebuild the access cache and we are planning to do
this over the coming weekend (18th & 19th February). This is a slow
and quite system intensive process, taking more than 24 hours, hence
needing to wait for a weekend to run it. The support teams have met
today (17th February) to finalise the details for this work. It is not
anticipated that customers will experience any slow running as a
result of this process, as we will work through this offline and
introduce the changes to the live site on Monday 20th.

3. The support team are investigating ways of improving the access
cache system to avoid this problem in the future. An initial piece of
this work has already been developed and is due for deployment in a
week’s time.

Once again, please accept our apologies for any inconvenience that has
been caused. Also thanks again to all in the library community who
have provided valuable feedback.

Kind regards
Cambridge Journals

Cambridge University Press
University Printing House
Shaftesbury Road
Cambridge, CB2 8BS
England, UK

http://www.journals.cambridge.org