Australian customers were left without access to the cloud-based SugarCRM customer relations management system for an hour on Thursday afternoon as the company conducted emergency maintenance work on a file server.
SugarCRM said it had scheduled the maintenance work on between 10pm and 11pm Wednesday Pacific time, which meant it took place at 4pm on Australia's east coast on Thursday.
However, a local SugarCRM representative told one customer that no advance warning was given of the maintenance.
Customers trying access the OnDemand system on Thursday saw an error message without further information pointing them to the company's Facebook and Twitter profiles to receive updated information, without providing an expected time of resolution.
However, SugarCRM provided no further information about the outage via Twitter.
The company's Facebook page told customers that the system's US-based data centre was down for scheduled maintenance for about 40 minutes.
SugarCRM, a competitor to Salesforce.com, provides CRM access through on-premises deployments, hosted access through local integrators as well as an 'OnDemand' software-as-a-service (SaaS) offering through a US-based public cloud.
It said earlier this year that it had 600 largely small business customers in Australia and New Zealand, evenly split over the three deployment methods.
The maintenance work meant the OnDemand instances of SugarCRM were unavailable.
Kevin McIsaac, of Sydney consultancy firm IBRS, said his organisation was not notified of the shut down and discovered only through SugarCRM's Facebook page that it was scheduled maintenance.
McIsaac said the schedule was a "bloody stupid time to do this". He added it was an example of SaaS immaturity.
McIsaac was also told by a local SugarCRM representative that all instances of the system's on-demand environment were unavailable yesterday, as work to migrate the global link repository to a new storage server took place.
Work was being conduct on an emergency basis "outside of the normal weekly maintenance window due to global performance problems currently occuring in OnDemand".
"I understand your reasons for being upset with the timing of the maintenance, although with Sugar being a global company, virtually any down-time would have affected some portion of our customers somewhere around the globe," the SugarCRM representative wrote in the email.
The representative confirmed no prior notice was given for the emergency work.
SugarCRM's global link repository is a directory with UNIX symbolic links that allow the web server process to resolve the correct path of an instance, back to its location on a file server.
The existing file server on which the repository was housed also ran other services and as a result, became "starved for I/O". This in turn resulted in performance for the file server declining "to an unacceptable point" for all SugarCRM OnDemand traffic as resolving the symbolic links took a long time.