I am running latest MacOS (10.12.6) and latest BOINC client (which I have downloaded and reinstalled as a result of this problem).

The BOINC client is exiting with this error. I cannot do anything in the client like add another client. The Climateprediction project is not listed....

Both Macs failed at the same time.

That suggests it is something wrong with my account or at the server end maybe?

Any suggestions?

Thanks

Simon

---

I meant "... like add another project..."

I can't even get the BOINC client to give me the version of the software. Most of the menu options are greyed out,

Simon

---

Look at the previous thread in this Mac forum. That may be what the problem is, and could hold the key to getting back into boinc. If that is what is happening, after you've fixed it, I would suspend the climateprediction.net experiment until such time as the work units causing the problem are removed or designated Windows only. An announcement will be made in this forum and the Linux one when that happens.

Thanks for the suggestion. But assuming I got it right that did not fix it. I looked in client_state.xml for those error messages but could not find them. So then I renamed those 2 files (client_state and client_state_prev) and restarted the BOINC client. But the issue persisted. I moved those 2 files to the desktop and repeated... same result.

Just to clarify. When I start the client, the message in the bottom right hand corner says it is "connecting to localhost". Then I get the "BOINC Client has exited" message, and the message in the bottom right corner then reads "Disconnected".

Thanks.

Regards

Simon

---

I'm not sure exactly as I don't run a Mac. Another possible solution is to delete the file account_climateprediction.net.xml from BOINC's data directory. If that doesn't do it, will have to wait for Iain, one of our moderators who does run a Mac, to come in and look at this.

On the assumption that your machine has not been affected by one of the recent failed long models (your machines are hidden, so I can't check that), then this looks like some sort of permissions problem at the BOINC Manager level. A complete reinstall might work if you delete all the BOINC folders after the uninstall; you mentioned a reinstall but not whether the old folders had been removed.

If BOINC Manager can't detect localhost after that then either the local network setup is completely corrupted (which would be affecting other things) or the new BOINC Manager version has some serious problem, in which case you could revert to an earlier version. I haven't upgraded my BOINC Manager because I only do that when I've run out of work, to avoid permissions problems, and I've still got some models to get through.

Looks like you have the same issue I had, please see my workaround

---

Thanks for the reply.

Happy to give that a whirl. But this happened on both my Macs ? Is the fix your are suggesting likely to address a problem that has independently affected 2 machines ?

Regards

Simon

---

Hello Simon,

I cannot be 100% positive as you stated that your 2 machines crashed at the same time. If you meant "exact same day, hour, min and second" it would be indeed strange and probably not related with what happened to me, but if you meant "same day" or "same hour", let's suppose that they were both affected by the recent failed CPDN long models the BOINC forums are talking about for the past few days.

If you are still stuck with Boinc, have a try on one mac. First backup your current "Boinc data" directory just in case, and then try the time machine restoration. You can even try to dig a litle in all the informations from the forums and just restore the appropriate files instead of the full directory. Iain may be able to provide more specific informations

---

Thanks. I'll try the Time Machine recovery over the weekend.

The two Macs suffered the same fate within a few days rather than minutes.

I have deleted all BOINC files and folders on one of the Macs, reinstalled BOINC. When I then restart the BOINC client I get the "Connecting to localhost".... it then goes to "The BOINC client has exited unexpectedly 3 times..." after a few minutes.

Ill keep you posted in the Time Machine restore

---

I restored the BOINC Data folder from my TimeMachine - and hey presto it all works again.
I have reset the project and Suspended i until further notice - assuming the problem is related to these dodgy work units.

I will apply the same process to my other Mac.

Thanks for the suggestions everyone

Regards

Simon

---

I have reset the project and Suspended i until further notice - assuming the problem is related to these dodgy work units.

The, "dodgy" work units have been deprecated for both Mac and Linux which means right now, very little work is coming for us except tasks that have already failed once or twice on other machines. I got two yesterday which promptly failed at the same point they had previously.

What I am not 100% sure about is whether any of the tasks that totally mess up BOINC could appear as resends for mac/Linux machines.

[quote]
What I am not 100% sure about is whether any of the tasks that totally mess up BOINC could appear as resends for mac/Linux machines.

It shouldn't since the WAH2 app is deprecated for Linux and Mac, and that app is what was used for the problematic work units. The crashes you recently had must have been hadcm3s?

The crashes you recently had must have been hadcm3s?

Yes, both had crashed previously on Windows machines at about 33seconds in cpu time.

I wsn't quite sure if the deprecating applied to retreads or just to tasks going out for the first time. Thanks for confirming that it is for all of them.