Thus said Philip Bennefall on Wed, 03 Sep 2014 21:50:28 +0200: > Was this a problem that could be said to be on the server side or the > client side?
The problem that was recently corrected was server side. Did one of you recently make a largish checkin (as in a large number of artifacts)? [longer explanation follows] Basically it happened when there were a large number of items in the unclustered table on the server that just arrived as part of a push from a large list of igots from a client. These artifacts were phantoms because the content had not yet arrived for them, but then the next sync operation caused the unclustered table to be cleaned up and it removed the phantom artifacts from the unclustered table. If one of those unclustered artifacts happened to be a checkin, your clients doing a sync wouldn't know about it because that artifact did not get included in any cluster artifacts. This missing checkin would remain missing until it had a descendent that was included in another checkin or artifact that was published in the servers list of igots. So, another potential way to cause it to sync (other than the --verily option) is to just have someone make another checkin (or perhaps add a tag or something maybe?) as a descendent of the missing artifact. Andy -- TAI64 timestamp: 4000000054078b5b _______________________________________________ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
