Envvars nodeprops #1
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
User can choose after build behavior of VM. 0. Suspend VM 1. Stop VM 2. Keep VM running 3. Return to previous state Instead "stopVm" now connector slave gets final command from VM.
…r-VM-stopping-mode Provide configuration for vm stopping mode
This is the first public plugin release on official Jenkins project hosting.
Fixed typo and changed repo URL.
As required for maven release.
Now plugin is hosted on jenkins-ci.org. Thus part of custom installation process is not needed. Also issue with incorrect computer launcher representation on the web form is fixed.
Add labels for host slave
If connector slave is not used as build node strategy "Demand" was set for it. This made possible such situations when connector slave was made offline due to inactivity for some time. And as a result command to stop virtual machine was not executed on connector.
Need to block restart untill all VMs are stopped or suspended or paused are required by their configuration. Bug #JENKINS-40628
Just don't create node object during provisioning if VM start failed or IP address was not obtained. Let Jenkins find better executor for this job. Bug #JENKINS-40685
Seems large VM on MacMini may resume for 2-3 minutes. Bug #JENKINS-40685
If VM slave node fails to start for some reason it will be left offline by Jenkins because we just skip check in retention strategy in such case for unclear reason. Bug #JENKINS-40685
Check for VM existence was implemented disfunctional in initial commit. Let's reimplement it. Bug #JENKINS-40685
Foundation stone of this procedure is proper errors handling. Previous attemt did not fix anything in fact. So that if, for example, Jenkins fails to connect to VM through SSH appropriate node stalls in offline mode with job handing on it. Bug #JENKINS-40685
To fix bug with proper limiting the number of VMs runnings on one host we need to rework VM start-up procesdure so that actual virtual machine start will be attempted only if it is not running already (if VM is already running on host no additional limiting could be applied to it). Bug #JENKINS-40685
We need to check resource limits (CPU and RAM for now as main resources) before VM start to avoid host overload. Otherwise in case of CPU or memroy overcommit host performance may degrade significantly, even to the state of unresponsibility. Yes, Parallels Desktop internally checks some limits before VM start, but: 0. these checks are not strict (this is bug or not -- I don't know); 1. for plugin these checks blocking looks like common VM start failure. So its better to distinguish resources checks issues with others to better handle such errors. Actual implementation goes in the next commit. Bug #JENKINS-40685
Here's general algorithm done for CPU counting and RAM separately: 0. take available CPU cores and physical RAM on host; 1. for each running VM sum virtual CPUs and RAM (main memory and video RAM); 2. add for each VM RAM 500 Mb as virtualization overhead; 3. subtract 1 Gb from host available RAM -- consider this to be taken by host OS, apps, Jenkins' java etc. 4. so available resources are: host CPUs minus sum of runnings virtual CPs, host RAM minus all VMs RAM; 5. compare available resources with start VM required resources, if CPU or RAM exceeds available values -- refuse to start VM. Bug #JENKINS-40685
At this point we didn't know if prlctl commant finished with sucess or or not, we just parsed its output. To improve error handling, particularly in case of resources exceeding cases, let's throw and catch appropriate exception if prlctl return code is not zero. Bug #JENKINS-40685
check() method of computer's retention strategy may be called from different threads. So parallel execution is possible and need to add locks into method code.
Since maven tests warn against using non-portable implementation from Oracle run-time.
check() method in retention strategy may be called after VM start but before launching slave on it. Plain isIdle() check returns true -- and we started disconnecting and removing node, _before_ initial connecting! So computer start fails and job is not executed. To fix this lets introduce idle timeout which should be passed before we start computer termination. As it is done in retention strategies of other plugins. 2 minutes value is "taken from the ceiling".
Original implementatino didn't work for multiple clouds configured in the system. Restart listenter worked only with one instance of the cloud class. So lets rework restart listener so that it will ask for restart for every PDfM host computer. Also need to count VMs for only _after_ they are started. Bug #JENKINS-40628
Stupid error in initial implementation was made: hosts CPUSs number and RAM size were calculated on the master side while they are expected to be taken from Cloud host slave. So need to add MasterToSlaveCallable for proper host resources determination. Bug #JENKINS-40685, improvement #JENKINS-40691
As suggested in #INFRA-588.
Previously we aborted VM computer launch if were unable to get its IP address via `prlctl list -f`. This could happen because Parallels Tools are not installed in that VM or just because of a bug in Parallels Desktop. But we may use a fallback way: just try to use address of hostname specified in a form of VM configuration. Bug #JENKINS-41431
At the moment I don't have valid environment to test and fix IPv6 support. So lets set up the plugin to work with v4 only for now. Bug #JENKINS-41431
Do not provision VM slaves when the host slave is offline. Update all VM slaves online/offline state in accordance with the host slave. Prohibit managing VM slaves connection from the GUI.
Implement 'Mark this node temporarily offline' for host slave
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.