There are a number of reasons that could cause ECFLOW ecFlow to stop working (server crashes, the computer ECFLOW ecFlow is running on crashes, etc.). The ecFlow checkpoint file allows ecFlow to restart at the point of the last checkpoint before a failure. This gives reasonable tolerance against failures.
When the server starts, if the checkpoint file exists and is readable and is complete, ECFLOW the ecFlow server recovers from that file. Once recovered the status of the server may not exactly reflect the real status of the suite, it could be up to a few minutes old. Tasks that were running may have now completed so the task status should be checked for consistency.
The checkpoint files can be read by any ecFlow running on any operating system
There are two separate checkpoint files. ECFCHECK
Code Block |
---|
ECF_CHECK ecf.check |
...
ECF_CHECKOLD ecf.check.b |
When ecFlow needs to write a checkpoint file it first moves (renames) the previous file ECF_CHECK to ECF_CHECKOLD and then creates a new file with the name ECF_CHECK. This means that you should always have a file that is good. If a crash happens while writing ECF_CHECK, you can still recover from ECF_CHECKOLD (by copying its contents to ECF_CHECK), although that version is not quite as up to date.
Info |
---|
Tip: You can copy the checkpoint files between systems. |
Another ecFlow server can be started with the original server's checkpoint file and take over from the original ecFlow server host in case of a catastrophic systems system failure.