[Kolab-devel] [issue687] Kolab2 kolabd sometimes deletes all deleteFlags without deleting main object

[Stephan Buys]

New submission from Stephan Buys <s.buys at codefusion.co.za>:

It might be worthwhile investigating the process around how deleteflags are 
cleared when slave servers cleanup Cyrus mailboxes. 
 
It is hard to replicate, but we have seen a couple of times a situation where 
the following happens: 
 
1) A user deletes a user on Kolab2 
2) a deleteFlag is set for each server in the organisation 
3) Slaves and Master start cleaning up mailboxes removing their deleteFlags as 
they continue. 
4) Due to timing/locking/unknown reasons the deleteFlags all get cleared but 
the user object in LDAP is not cleared. 
- I assume that it is up to the last server in line to delete the object. 
5) The user now appears "undeleted" in the Kolab admin interface (all 
deleteFlags removed) 
 
It could take several retries to get the object deleted, everything has to 
happen in concert. 
 
I suggest that rather than removing the deleteFlags an extra attribute is set 
that flags the status of the Cyrus deletion from each server. This way a 
daemon wont recreate a mailbox accidently. (deleteFlags never get removed). 
 
Then when all servers indicate that the Cyrus mailboxes have been cleared the 
master server goes and removes the object from LDAP. 
 
At the very least some mechanism should be in place to assure that a object 
can never be "undeleted" by the removal of deleteFlags.

----------
messages: 4091
nosy: bh, steffen, stephanb
priority: bug
status: unread
title: Kolab2 kolabd sometimes deletes all deleteFlags without deleting main object
topic: server
________________________________________________
Kolab issue tracker <kolab-issues at intevation.de>
<https://intevation.de/roundup/kolab/issue687>
________________________________________________