[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: slurpd blocking updates

--On Tuesday, May 11, 2004 8:27 PM +0200 Pierangelo Masarati <ando@sys-net.it> wrote:


I am again seeing slurpd failing to update replica's under 2.2.11.  Is
anyone else seeing this?

cat slurpd.status

slurpd has failed to even talk to ldap-test2 since May 9th, yet it
updates  ldap-test1 and ldap-test3 regularly.  ldap-test2 is fully
responsive to  queries, and slapd is restarted on all systems daily.
slurpd is restarted  on the master daily.

I think I've seen something similar recently: I started a master, a slave and slurpd simultaneously; the I added an entire database via ldapadd (few hundreds of entries); the master was populated correctly, and slurpd moved the master's replog into its, but the slave didn't receive anything. As soon as I noticed it (few minutes later), I restarted everything, and the replication occurred instantaneously. I was using HEAD with bdb on both servers, so it's not that different from 2.2.11. I couldn't reproduce it, so I really have no clue. Since I was trying to quickly improvise something totally unrelated (playing with Samba 3 as BDC acting on a couple of slave/master DSA), I'm not even sure I did everything correctly from the beginning. I'll keep my eyes open to see if anything similar occurs.

Can you suggest guidelines (at least hints) to modify test006 in order to
try to induce your problem?

For me, I can get the behavior to happen fairly regularly just by sending a group of changes. Last night, I did 7 deletes (which went slowly on my 2.2.11 test master, and quickly on my 2.2.6 production master). I've hit this problem before when I did a series of:

Delete 50 existing entries
Add the 50 entries with different data


Quanah Gibson-Mount
Principal Software Developer
ITSS/TSS/Computing Systems
ITSS/TSS/Infrastructure Operations
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html