Peter Palmreuther wrote:
hyc@symas.com wrote:
Peter Palmreuther wrote:
hyc@symas.com wrote:
[...[
Thanks. Please try the syncprov.c in CVS HEAD. (Also you should probably be testing the RE24 code now, not 2.4.19.)
[...] ran several thousand test iterations without a single failure yet. [...] Is this fix "back-portable" to 2.4.19 source, so I can compile a 2.4.19-custom version??? [...]
The patch will be in 2.4.20, which will be released soon.
Thanks a lot :-)
Just for the record, so I can
- close this case internally, and
- justify the upcoming upgrade toward our change and release management:
What was the problem? The second server sending back received synced operations or the first server processing those "send backs"?
Both. The consumer was sending the wrong SID to the provider, so the provider's checks to prevent "send backs" wasn't catching it. Also the provider's checks were being bypassed on Modify operations, so the modified entry would be sent back and recreated after it was deleted. I've added a sequence of repeated add-modify-deletes to test050 to make sure we don't regress on this in the future.
Or was it just "some pixie dust" on the WLAN cable? ;-)