[ https://issues.jboss.org/browse/JBIDE-18862?page=com.atlassian.jira.plugi... ] Darryl Miles edited comment on JBIDE-18862 at 1/21/15 7:37 AM: --------------------------------------------------------------- If you are checking for *.isdeploying (which is also a server managed marker file). Then yes you can remove a *.dodeploy marker that still exists from a previous publish that has not yet been actioned by the AS. Multiple sequential publishes in the Eclipse are normal, the first might take 10 seconds to project build and publish the next 2 publish operations (due to editor file saves) might take 1 second for each to complete. It is normal for the AS to not notice the *.dodeploy yet. If the AS did notice then the publish operation (and has started to deploy) Eclipse should be blocking due to the *.isdeploying marker being down. Since it should not be performing any file change operations in the deployment/ directory until there *.isdeploying marker has been removed by the AS. File changes can be made in Stopped and Started states, but not in Stopping and Starting states. There is no notion of Full Publish or Incremental Publish by the AS. This is an Eclipse'ism. There is only the notion that Eclipse has a target runtime state and the AS has a current runtime state and actions to perform AS state changes. So yes... an incremental publish that removed the marker before making any file change operations should at the end of the operation review the situation. It will find that Eclipse has a target runtime state to achieve of "AS Started" and "Module Started". It will note that Eclipse is already started and take no action. It will note that the Eclipse module is not started (or is configured to be restarted on change) and signal the appropriate action to the AS to bring the runtime AS state into alignment with the Eclipse target runtime state. Once Eclipse target state matches the AS runtime actual state, we have achieved a nominal state. was (Author: dlmiles): If you are checking for *.isdeploying (which is also a server managed marker file). Then yes you can remove a *.dodeploy marker that still exists from a previous publish that has not yet been actioned by the AS. Multiple sequential publishes in the Eclipse are normal, the first might take 10 seconds to project build and publish the next 2 publish operations (due to editor file saves) might take 1 second for each to complete. It is normal for the AS to not notice the *.dodeploy yet. If the AS did notice then the publish operation (and has started to deploy) Eclipse should be blocking due to the *.isdeploying marker being down. Since it should not be performing any file change operations in the deployment/ directory until there *.isdeploying marker has been removed by the AS. File changes can be made in Stopped and Started states, but not in Stopping and Starting states. There is no notion of Full Publish or Incremental Publish by the AS. This is an Eclipse'ism. There is only the notion that Eclipse has a target runtime state and the AS has a current runtime state and actions to perform. So yes... an incremental publish that removed the marker before making any file change operations should at the end of the operation review the situation. It will find that Eclipse has a target runtime state to achieve of "AS Started" and "Module Started". It will note that Eclipse is already started and take no action. It will note that the Eclipse module is not started (or is configured to be restarted on change) and signal the appropriate action to the AS to bring the runtime AS state into alignment with the Eclipse target runtime state. Once Eclipse target state matches the AS runtime actual state, we have achieved a nominal state. -- This message was sent by Atlassian JIRA (v6.3.11#6341)