The plugin provides flexible disk selection strategies. Currently, the user may choose to allocate a Disk from the Disk Pool using one of the following strategies:
- Select the Disk with the most usable space
- Select the Disk with the highest read speed
- Select the Disk with the highest write speed
The default allocation strategy can provided in the Jenkins global config, for each Disk Pool entry, as shown in the image bellow:
The disk allocation strategy can then be overridden in the Pipeline code, as we will see in the afferent sections.
This strategy allocates the Disk that has the most usable space within the Disk Pool. The Pipeline code can override the global config as following:
def extWorkspace = exwsAllocate diskPoolId: 'diskpool1', strategy: mostUsableSpace()
// ...The strategy also accepts an optional parameter: estimatedWorkspaceSize.
This parameter ensures that the allocated Disk has the usable space greater that the estimated workspace size
(disk's usable space > estimated workspace size):
def extWorkspace = exwsAllocate diskPoolId: 'diskpool1',
strategy: mostUsableSpace(estimatedWorkspaceSize: 100)
// ...In our case, we have provided an estimated workspace size of 100 MB. The value has to be in MB. Let's assume that the allocation strategy has selected the disk with the most usable space, that has a value of 80 MB. The disk's usable space is smaller than the estimated workspace size. The job will fail, because it could not find any disk that has the usable space > estimated workspace size.
This strategy uses the disks' read speed to allocate the proper disk. Currently, the only way to know this information is if we provide it in the Jenkins global config. In the External Workspace Definitions section, for each Disk, there is the Disk Information entry:
After we have set the custom Disk Information, we can proceed forward to specify the allocation strategy that is based either on the disk's read speed, or on the disk's write speed. The allocation strategies can be set in the Jenkins global config (as we have seen before), or it can be overridden in the Pipeline code:
def extWorkspace = exwsAllocate diskPoolId: 'diskpool1', strategy: fastestReadSpeed() Similar as in the previous example, we can specify the estimatedWorkspaceSize parameter:
def extWorkspace = exwsAllocate diskPoolId: 'diskpool1',
strategy: fastestReadSpeed(estimatedWorkspaceSize: 100) If this parameter is provided, the strategy will alocate the disk with the fastest read speed, but that has its usable space at least equal to the estimated workspace size.
For example, let's assume that we have disk1 with read speed of 25 MB/s and usable space of 120 MB, and disk2 with its reading speed of 30 MB/s but with a smaller usable space, 80 MB.
If the estimatedWorkspaceSize parameter is given, with a value of 100 MB, the allocation strategy will select disk1, because it has the minimum required usable space, even though its read speed is smaller than disk2.
If we do not pass the estimatedWorkspaceSize param, the strategy will ignore the disks' usable space and will allocate disk2, because it has the highest read speed.
This strategy is very similar with the previous one, the main difference being that the disk is selected based on its write speed. For this, the disks' write speed has to be provided in the Disk Information section.
In the Pipeline code, it can be overridden as:
def extWorkspace = exwsAllocate diskPoolId: 'diskpool1', strategy: fastestWriteSpeed() Or, with the estimatedWorkspaceSize parameter:
def extWorkspace = exwsAllocate diskPoolId: 'diskpool1',
strategy: fastestWriteSpeed(estimatedWorkspaceSize: 100)Same principles apply as the previous selection strategy.