Header Search Widget

IRIS

Instructional & Research Information Systems

Sep 2013 Server Patching

by

Microsoft will release their patches for the month on Tuesday, Sep 10th, 2013.

PRINT and WINTERM will automatically patch and reboot Wednesday, Sep 11th, at 3:00 am.

WINTERM may be unavailable till 10:00am Wednesday, Sep 11th. This depends on the needed level of application updates in addition to the operating system updates.

Resolved as of 2013-09-23 11:55:00

filesystem problem in LDAP server may have caused temporary authentication failures

by

One of the LDAP servers in our pair became unresponsive due to a filesystem problem. Some services (such as Jabber) had authentication problems during this time. The system is back online now after running fsck.

The current LDAP servers are old legacy systems slated to be replaced in the very near future. The configuration of these servers and the services that rely on it is non-optimal which results in problems with authentication to some (but not all) services that rely on LDAP.

The new LDAP system is in its final testing phases now and will completely replace the aging infrastructure we have now.

The problem with the system was detected by IT staff with one of our monitoring systems which alerted us. We are not aware of any impact to the community and have received no reports from anyone. If you experienced any authentication problems between 1-2am then please let us know by writing to help@eecs.berkeley.edu

Resolved as of 2013-09-07 01:57:00

New LDAP Infrastructure Ready for Testing

by

IRIS is pleased to announce that our new LDAP infrastructure is ready for public testing. The new LDAP servers are now accessible from any on-campus host at openldap.eecs.berkeley.edu. TLS is supported on port 389 and SSL on port 636, using a new certificate from InCommon.

#Testing
Programmers and system administrators are encouraged to test any applications or scripts they maintain against this new infrastructure. The only change that should be needed for testing is to change the hostname to openldap.eecs.berkeley.edu.

Please send any questions or report any problems that come up during testing to [help@eecs.berkeley.edu](mailto:help@eecs.berkeley.edu)

#Architecture
openldap.eecs points to a redundant load balancer, which at this time is routing ldap requests to one of three backend OpenLDAP servers that are running in multi-master mode. These load balancers and backend servers are a combination of physical and virtual hosts located in independent data centers across campus (Sutardja-Dai Hall, Soda Hall, Warren Hall). With this architecture, a power or network failure in any one building will not bring the service down.

#Schema Changes
The schema is largely unchanged, save the removal of objectclasses and attributes that were added by the Sun/iPlanet IMAP and Calendar applications previously in use in the department. With the migration to bMail and bCal, these attributes are no longer needed and so have been removed.

#Timeline

  • 9/6 openldap.eecs initial public testing with a snapshot of test data. Reliability testing ongoing. [1]
  • 9/11 openldap.eecs is syncing real data from ldap.eecs. firewall is opened up to world. Testing continues. Reliability testing done.
  • 9/18 Assuming no major problems reported, ldap.eecs and ldap.cs become nicknames for opendap.eecs

#Notes
[1] During the next week IRIS will be conducting reliability testing, meaning we will be taking down some of the servers to ensure the service as a whole stays up. This testing might affect performance but should be completed by 9/11/2013.

AirBears2 now available in EECS

by

AirBears2 is now being broadcast in EECS airspace*. Connecting to AirBears2 will require that you create an AirBears2 key. For more information, including steps for generating a key, visit http://airbears.berkeley.edu. If you have any trouble, please contact the helpdesk at help@eecs.berkeley.edu.

Thanks,
IRIS IDSG Network Group

*EECS airspace includes Soda, Cory, Sutardja Dai, Calvin, and Blum Halls, as well as the Sibley Auditorium, Hearst Memorial Mining Building, and BWRC.

Soda 4th floor switch hardware failure

by

A hardware failure of a network switch has affected part (but not all) of the wired network on the 4th floor of Soda Hall.

We are working to replace the failed switch, but if you need your device re-connected sooner you can contact the helpdesk at help@eecs or call 510-642-7777. Provide your device hostname, MAC address, and the port cable ID where your network cable is plugged into the wall, and we’ll move your device to another port so you will have service restored while we continue working on the failed switch.
[Read more…] about Soda 4th floor switch hardware failure

Electrical work for UPS in SDH related to cluster.eecs and inst-fs storage

by

On Friday, as well as continuing on Monday and Tuesday (Aug 23-27) there will be electric work performed to hook up a UPS for the cluster.eecs and inst-fs.eecs storage devices. We do not anticipate any interruption of service and no downtime is planned, however we want to make you aware of the work being done.

Resolved as of 2013-08-27 19:42:00

cluster.eecs ran out of inodes

by

cluster.eecs ran out of inodes at 4:12am. We’ve increased the number of inodes by 20%. To prevent this problem in the future we will monitor the inode usage in addition to the space usage and adjust the inodes available before it runs out again.

Resolved as of 2013-08-16 06:30:00