Azure roles are cycling - WaHostBootstrapper.exe crashing, faulting module ntdll.dll -

this driving me crazy: have windows azure cloud service asp.net mvc 3 project. there have been many changes lately, have deploy every few days.

sometimes (e.g. now), we're stuck on following deployment error:

role instances recycled amount of times during update or upgrade operation...  

manage.windowsazure.com tells instance 0 of our staging deployment non-healthy:

recycling (role has encountered error , has stopped) 

instance 1 fine, web services on instance 1 working, so cause of error not our code.

we rdp'ed instance , examined application event log, displays following error multiple times:

faulting application name: wahostbootstrapper.exe, version: 6.0.6002.18488, time stamp: 0x505cf7da faulting module name: ntdll.dll, version: 6.1.7601.17696, time stamp: 0x4e8147f0 exception code: 0xc0000008 fault offset: 0x00000000000d4995 faulting process id: 0x970 faulting application start time: 0x01ce474976d706d2 faulting application path: e:\base\x64\wahostbootstrapper.exe faulting module path: d:\windows\system32\ntdll.dll report id: c26d8be8-b33c-11e2-a9be-00155d3ab8c9 

when happens, are

• re-imaging instance
• re-booting instance
• deploying again
• creating support ticket

some hours later, works again. we didn't change in deployment, works again, did before. @ time microsoft support reacting, fine again, cannot show them problem.

this ridiculous , frustrating. we're losing days of work because of silly error.

anyone else having these problems? ideas how stop that?

check following:
1. have latest azure sdk , solution using correct dll references. e.g. make sure if have v2.0, references use same version. if not, build afresh.
2. references (including packages , own libraries) enabled "copy local" = true , included in package bin.
3. if using storage, check connection strings , validity of keys.
4. check whether service configuration has correct osfamily , osversion targeting.
5. if nothing helps, try intellitrace.