Help and FAQ

Possible runaway thread

When this alert is generated, it is simply meaning that the named thread has taken nearly 100% cpu on a single core. This can be for many reasons:

  1. A client application is reusing the same connection with little to no idle time, and is normal;
  2. A query with a very large result-set is streaming the result for more than a minute;
  3. A backup is being made via a connection, and is saturating a single core;
  4. Related to #2, but possibly a malicious user has gained access to the system, and is downloading lots of data;
  5. Also related, but if the proxy is open to the internet, someone may have hacked an account, and is downloading all the data from the database;
  6. There is a bug in Heimdall, and a thread is in an infinite loop.

If the thread name contains an IP address, this is a thread being used to service a connection, and likely fits one of the first five cases. If it doesn't contain an IP, please provide logs and a screen dump of the alert for analysis by Heimdall support.