In recent months I’ve been mainly focused on Hyper-V installations, but I’ve had an interesting VMware week this week. After 7 years of using ESX and then ESXi in production i experienced my first Purple Screen of Death which although inconvenient at the time i don’t think is too bad going!
Then later in the week and the subject of this post vCenter stopped running on one of our oldest installations.
On checking the event logs there were two event errors that were repeated a number of times, event id 1827, and event id 1105.
The installation in question is running vSphere 5.1 and was setup using the bundled SQL Server 2005 Express, which has a 4GB limit. So from the event logs above it’s pretty clear that the issue is the database is at or near capacity. After doing a little bit of research online I found the following two VMware knowledge base articles.
The basic process to resolve the issue is to stop vCenter (if it is running), download and run the provided script within SQL Management Studio to purge the database by specifying the number of days beyond which you want to remove the events. i.e purge items older than 3 years by specifying 1095 days.
The KB suggests purging a small amount of data at a time as the script can take a while to run. The database in question had events and alarms dating back over 4 years, so i removed them a year at a time, and each time I ran the script it took between 3 – 5 minutes.
Ultimately I left a years worth of data in the database which reduced the size from 4GB to 2.8GB, as I plan to upgrade it to a later version of SQL in the near future.
By the way, the Purple Screen of Death was caused by an unhappy E1000 driver and an ESXi server which wasn’t on the latest update, all patched and running happily again now!