2

Events till now

We have a CC 7.1.2.2, multisite setup where we do deliveries between 2 sites. Now when resuming a delivery at the destination site, we get this error :

Unable to read change set entry for activity "<activity name>". Unable to 
convert diffs to elements. Unexpected error in deliver. Unable to perform merge. 
Unable to do integration.

Then running checkvob -ucm shows some broken hyperliniks which the SCM support team fixes for us. IBM tech note says this is a synch issue.

Now the actual problem:

This has started happening on a regular basis suddenly and we know its NOT a synch issue between VOB and PVOB as the packets are getting synched properly. What I am interested in finding out is whether this could occur due to a specific set of user actions like deleting checked out versions etc. The key point is its not a one off thing and impacts our deliveries everyday. We are not able to find any concrete triggering actions or root cause

Any ideas ?

Machavity
  • 30,841
  • 27
  • 92
  • 100
Pulak Agrawal
  • 2,481
  • 4
  • 25
  • 49
  • Also, please note that we do not have `rmver` or `delete` rights.. all we can do is checkout the parent directory , remove the element and check in the parent effectively making the element invisible in the view, not really deleting it, so the post here > [link](http://blog.naver.com/PostView.nhn?blogId=sonjae76&logNo=10025866628&parentCategoryNo=2&viewDate=&currentPage=1&listtype=0) also does not help – Pulak Agrawal Nov 17 '11 at 05:09

2 Answers2

2

This has been linked to a multi-site sync issue from a long time now (here an example in 2005), and was also associated with a bug in CC multi-site 7.0.

But if you are really sure multi-site sync is not the issue, then it could be linked to "lost+found" issue, where elements could have been:

  • deleted (rmelem by an admin -- I know regular users in your setup don't have rmver or rmelem rights -- in order to clean the lost+found directory automatically, maybe through a ClearCase scheduled job or some kind of trigger?)
  • not selected because the config spec of views involved by your deliver are setup to not select the lost+found directory
VonC
  • 1,262,500
  • 529
  • 4,410
  • 5,250
  • thanks but : 1. the bug you've mentioned is when deliver -complete fails, while in my case deliver -resume is what is failing. 2. The bug as per IBM should be resolved in my version (7.1.2.2) 3. Lost+found is a pointer, I'll check with the admins 4. Nothing has changed for our views since months and so I don't think config spec has anything to do here. – Pulak Agrawal Nov 17 '11 at 07:36
  • @PulakAgrawal: 1. and 2. I am aware of your points, I was only mentioning those examples to illustrate how sync issues were involved for a long time, even if you say they are not in your case. 3. ok. 4. Just check in your config specs used by your delivers if there is no "`element /yourVob/lost+found -none`" rule. – VonC Nov 17 '11 at 07:41
  • Checked 4. No config spec rules like that, but still finding with the admin whether they have a job which purges lost+found. Will respond tomorrow on this. Thanks – Pulak Agrawal Nov 17 '11 at 10:16
2

Found out the issue; it was hidden synch issues indeed. What really was happening was that multisite synch was timing out for packets larger than 250 meg. This would create problems for bid inetersite deliveries where PVOB would synch over and VOBs would not. This was hidden as otherwise sync was happening properly.

Thanks VonC for the other inputs; I know you'd have pointed me to synch issues as a first measure had I not confirmed it wasn't the issue.

Pulak Agrawal
  • 2,481
  • 4
  • 25
  • 49