Re: (ITS#4088) hdb deadlock between syncrepl/slapcat

[hyc@symas.com]

richton@nbcs.rutgers.edu wrote:
> Full_Name: Aaron Richton
> Version: 2.3.11
> OS: Solaris 9
> URL: ftp://ftp.openldap.org/incoming/richton-051014-debuginfo.txt
> Submission from: (NULL) (67.85.180.27)
>
>
> Kicking the tires on 2.3.11 on a test machine. I've got a slapd doing syncrepl
> (consumer), and a slapcat, that appear deadlocked:
>
> slapd/6:lwp_cond_wait(0xFDE492F0, 0xFDE492D8, 0x00000000) (sleeping...)     
> slapcat:lwp_cond_wait(0xFE8FBB30, 0xFE8FBB18, 0x00000000) (sleeping...)
>
> I don't want to spam a 15k e-mail out; see FTP URL for debug traces.
>
>   
Interesting. Your current slapcat is using locker ID ec, the page in 
conflict is held by locker ID ef. A slapcat process only uses a single 
cursor to dump the id2entry DB so it probably is not the owner of ef. 
Locker ef has no other locks and is not waiting for any other locks; my 
guess is that it's a stale lock from a previously aborted slapcat run. I 
don't see why any part of the slapd task would have acquired that lock. 
Unfortunately BDB doesn't record process or thread IDs with locker IDs 
so there's no quick way to find the culprit.

-- 
  -- Howard Chu
  Chief Architect, Symas Corp.  http://www.symas.com
  Director, Highland Sun        http://highlandsun.com/hyc
  OpenLDAP Core Team            http://www.openldap.org/project/