I run Boinc +seti+cpdn on a bi xeon HT (means 4 proc), RHE4 64 bits. All cpdn clients crash when boinc pause them in order to run cpu bench. cpdn clients are paused, removed from memory (boinc prefs sais the contrary: leave in memory...), then exit because \"got signal 11\".
They are registered as \"computation error\". The result is that I cannot send more than 4 to 6 tricks for each model. What is wrong ? Boinc is 5.2.13 version, cpdn is running sulphur models.
I think I have the same problem on a Macos10 system (bi G4 proc), I have still to verify that the crash occurs because of the cpu bench.

---

Hi,

Sorry to hear this. Ever since 5.2.x and with the preference setting \"Leave applications in memory when preempted\" set to yes, I\'ve had no signal 11 errors. When a benchmark is done, it clearly states on mine, \"Pausing result resultname (left in memory)\". Do you have more than one set of preferences (work, home, school) and perhaps one of those sets has the leave in memory setting set to no?

Hi,

Sorry to hear this. Ever since 5.2.x and with the preference setting \"Leave applications in memory when preempted\" set to yes, I\'ve had no signal 11 errors. When a benchmark is done, it clearly states on mine, \"Pausing result resultname (left in memory)\". Do you have more than one set of preferences (work, home, school) and perhaps one of those sets has the leave in memory setting set to no?

I erase then reinstall boinc on that computer. That will take a few days before a crash, if any. Concerning the preferences sets, I specified only one, and no particular sets for home/school/work. But I noticed that the faulty computer had \"home\" in the location field. Looking at others ( I run around 10 computers), some have nothing, some have home. All of them are running smoothly, but they all have different configs ( 1 or 2 cpus, macos or linux, running 1 or 2 projects). The only one similar to the problematic computer has 2 cpus, fedora 32 bits, and runs seti+cpdn. It is running boinc 4.19 without any problem since a lot of time. Also its location field is empty. Then, wait and see. (at least wait for the next cpu test).

---