He would probably be just as happy if you wrote him and told him about how use 
used Process Monitor and the problem you solved with it. He does a presentation 
at TechEd every year called "The Case of the Unexplained…" where he talks about 
interesting cases he has found or people have sent him. He always ends them 
with "If you’ve solved one, send me a description, screenshots and log files".

…Tim

From: [email protected] [mailto:[email protected]] On 
Behalf Of Richard Stovall
Sent: Monday, June 02, 2014 9:40 AM
To: [email protected]
Subject: Re: [NTSysADM] CRM 2011 SSL cert error - Completely down

Aboslutely.  I'll probably never meet Mr. Russinovich in person.  If someone 
would buy him an adult malt beverage of his choice and send me the bill, that 
would be much appreciated.

On Mon, Jun 2, 2014 at 11:16 AM, 
<[email protected]<mailto:[email protected]>> wrote:
Process Monitor rules.

You should blog it up, looks like another interesting event where SysInternals 
saves the day :-)
Despatched via Blackberry. Mock if you will, but it gets my email without a 
fuss.
________________________________
From: Richard Stovall <[email protected]<mailto:[email protected]>>
Sender: [email protected]<mailto:[email protected]>
Date: Mon, 2 Jun 2014 11:04:44 -0400
To: <[email protected]<mailto:[email protected]>>
ReplyTo: [email protected]<mailto:[email protected]>
Subject: Re: [NTSysADM] CRM 2011 SSL cert error - Completely down

This was the third time I've installed a wildcard cert on this server in its 
2.5 year life.  Never had a problem before.  This time, obviously, there were 
problems.  The AppPool's account could not read new cert's private key.  All of 
the research I did said to manually add a read ACE on the private key's ACL in 
the certificates mmc snapin.  I did that with no effect.  Some others people 
reported success by subsequently adding a similar ACE to the key's ACL in the 
machinekeys folder.  That did not help either.  I worked on it for several 
hours yesterday before calling PSS.  When they did call me back, two gentlemen 
worked on it for about 7 hours and got nowhere.

While they were working on it, one of the techs added a read ACE for the 
AppPool's account to the machinekeys folder's ACL so that the permission would 
propagate down, but that did not help.  Soon thereafter the CRM guys decided 
that it was definitely a Windows issue and punted to another team.  This 
morning, while waiting for the second team to call, I fired up procmon to see 
what the heck was really going on.  It turned out that the w3wp process was 
trying to access a key in the machinekeys folder that did not exist anywhere on 
the drive.  Seriously.  Phantom stuff.  I e-mailed a screenshot to the CRM PSS 
techs and they suggested deleting the cert and re-importing it.  I did and 
presto!  Back in business.

It turns out that the ACE on the machinekeys folder's ACL has to exist BEFORE 
you import the certificate or the whole thing is mangled, and mysterious calls 
to non-existent keys are created.  Why this was not an issue the prior two 
times I installed previous iterations of the same certificate on the same 
server is unclear, nor is it evident why the (undocumented) procedure has to be 
in the correct (undocumented) order.

I hope none of you ever have to deal with this, but if you do, the fix is 
really simple.  (If you know the secret handshake.)



On Mon, Jun 2, 2014 at 9:55 AM, Richard Stovall 
<[email protected]<mailto:[email protected]>> wrote:
Yup.  Very.

It is finally resolved.  I'll write a post-mortem in a bit.

On Mon, Jun 2, 2014 at 8:18 AM, Michael B. Smith 
<[email protected]<mailto:[email protected]>> wrote:
That's....odd.

Sent from my Windows Phone
________________________________
From: Richard Stovall<mailto:[email protected]>
Sent: ‎6/‎2/‎2014 6:42 AM
To: [email protected]<mailto:[email protected]>
Subject: Re: [NTSysADM] CRM 2011 SSL cert error - Completely down
Still no resolution.  The CRM people are handing it to the Windows team because 
they believe it is an issue with the IIS AppPool not being able to read the 
private key.  (Which is what it looked like all along.)  None of the known 
fixes are working, however.  Ugh.

On Sun, Jun 1, 2014 at 10:09 PM, Richard Stovall 
<[email protected]<mailto:[email protected]>> wrote:
Yep.  I have done this before without trouble, but this time it's not working.

I did get a call from PSS (apparently there's one CRM engineer on duty on 
weekends) about an hour ago and he's (very politely) scratching his head and 
having me repeat all the troubleshooting steps I've already gone through.  
Thanks for checking.  If we ever get this resolved, I'll post the fix.

On Sun, Jun 1, 2014 at 9:50 PM, Susan Bradley 
<[email protected]<mailto:[email protected]>> wrote:
Have you reran then claims config wizard and ifd config wizard after replacing 
the SSL cert ?

(passed this onto a crm listserve)


On 6/1/2014 11:01 AM, Richard Stovall wrote:
Any one out there using CRM 2011 on premise?  I have been for years and went to 
replace an expiring SSL cert today and now cannot successfully enable claims or 
IFD.  Getting the dreaded "keyset does not exist" error.  I have granted the 
network service account access to the private key, which is usually the answer, 
but this does not work in this case.

I'm completely stumped and PSS is apparently not available to work on CRM 
issues on weekends.







Reply via email to