If you think there is a problem or a missing feature in the wrapper app, please report the issue to the BOINC dev team.

the Feature is there.

so you are using

checkpoint_filename

inside your wrapper - script and use it?

Thanks for pointing this out. We will add it to the config.

It is fixed in the latest app version: 220606.

Christian you need checkpoints. Otherwise people are wasting their time & money.

Christian you need checkpoints. Otherwise people are wasting their time & money.

??
My wus are restarting correctly after pc reboot...

DaveW: checkpoints are working. Maybe you still had an old app version?

I still question the usefulness of the checkpointing....the check points seem to happen near the start of the task and then stop for hours. Thus when used many hours of crunching time is still lost. (Tasks from 2206.11)

Hi Werinbert, checkpoints are written everytime the progress is updated. Similarly to your observations, we noticed issues with long-running tasks. Therefore we have reduced the taks size to 50% in 220612. We hope that this will reduce such issues. Thanks for reporting it.

Hi Werinbert, checkpoints are written everytime the progress is updated. Similarly to your observations, we noticed issues with long-running tasks. Therefore we have reduced the taks size to 50% in 220612. We hope that this will reduce such issues. Thanks for reporting it.

I just aborted a 2206.10 task that had run for almost 19 hours and was only about 50% complete, hopefully others do much better. I am still running a 2206.10 task and then all the rest are 2206.11 tasks, 39 of them on this machine alone.

Hi Werinbert, checkpoints are written everytime the progress is updated. Similarly to your observations, we noticed issues with long-running tasks. Therefore we have reduced the taks size to 50% in 220612. We hope that this will reduce such issues. Thanks for reporting it.

I just aborted a 2206.10 task that had run for almost 19 hours and was only about 50% complete, hopefully others do much better. I am still running a 2206.10 task and then all the rest are 2206.11 tasks, 39 of them on this machine alone.

Dump them mikey,
I dumped all my 2206.10 due to excess run time limits (I think I only managed 4 successful results), skipped version 2206.11 and now have 2206.12 which have a reduced run time so don't hit the exceed limit and I have not had any failures so far with 20 odd already run.

Conan

Hi Werinbert, checkpoints are written everytime the progress is updated. Similarly to your observations, we noticed issues with long-running tasks. Therefore we have reduced the taks size to 50% in 220612. We hope that this will reduce such issues. Thanks for reporting it.

I have recently seen a task that has been running for more than 5 hours without a checkpoint being written.

The idea of checkpoints is that they are written at regular periods, the default BOINC option is set to 60 seconds which always seemed to have been set quite low but can be configured by the user.
Could you write an entry in the logs to show when it is done ?
Assuming you can also fix that the entirety of each log is included as per my other post on that matter.

Our stats confirm that the average compute time is around 2 CPU hours. There can be very rare / exceptional cases where it takes much longer. If you see this happening, you can send us the WU to investigate it further.