OpenLDAP 2.4

["Allan E. Johannesen" <aej@WPI.EDU>]

>>>>> "hyc" == Howard Chu <hyc@symas.com> writes:

hyc> The entryCSNs used for replication now record timestamps with microsecond
hyc> resolution, instead of just seconds.

As slapcat'd from 2.3.39, I have CSNs in various forms:

grep CSN: users.ldapv2.2007.11.08.ldif |head
entryCSN: 20071030125340Z#000000#00#000000
contextCSN: 20071108000208Z#000000#00#000000
entryCSN: 2002120818:17:53Z#0x0061#0#0000
entryCSN: 20060412175546Z#000001#00#000000
entryCSN: 20070701134407Z#000000#00#000000
entryCSN: 20060412175617Z#00000c#00#000000
entryCSN: 20070911105027Z#000002#00#000000
entryCSN: 20070908105022Z#000000#00#000000
entryCSN: 20060412175733Z#00000f#00#000000
entryCSN: 20070531105422Z#000000#00#000000

The insertion into 2.4 fails:

slapadd: schema_init.c:3486: csnNormalize: Assertion `val->bv_len == (sizeof("YYYYmmddHHMMSS.uuuuuuZ#SSSSSS#SID#ssssss")-1)' failed.
Abort

My guess is that I need to warp those CSNs into the new form before I can load
my data.

Is there a script I've overlooked that accomplishes that?

Thanks.

Hmmm.  Looking at the code as I type this, perhaps the only evil one is the one
with colons and the "0x" hex field.  If I change

entryCSN: 2002120818:17:53Z#0x0061#0#0000

to

entryCSN: 20021208181753Z#000061#00#000000

will that be sufficient to let the data load?  Mercifully, I only have 67 of
those in my small database.

Thanks again.