|
From the documentation
Head Disk Type Indicates whether the disk type is SSD or HDD. If you change a daemon to SSD, then jobs with an oplog greater than 1GB/hour will go to SSDs and those with an oplog less than 1GB/hour will go to HDDs.
Q: So what happens when the customer has a job that doesn't have an oplog churn of at least 1GB per hour and only one daemon, an SSD daemon?
A: The job will sit there forever, unclaimed, and not happening.
This is fixed in 3.4, but still an issue in 2.0.3, the yet to be released 2.0.4, and older versions of Ops Manager. We need to change the docs to make this implication more clear.
My suggested rewording is
Indicates whether the disk type is SSD or HDD. If you change a daemon to SSD, then only jobs with an oplog greater than 1GB/hour will go to this daemon. Any jobs with an oplog less than 1GB/hour can only go to a daemon marked HDD.
|