On 6/3/08, stack <[EMAIL PROTECTED]> wrote: > Does tupdatesLock prevent the hole you describe? Flushing must hold the > updatesLock write lock to proceed. Updates take out a read lock on > updatesLock. All updates will have gone into the WAL and into the memcache > before the read lock is released.
Do you mean the updatesLock's write lock acquired before the calls to HStore.snapshot()? But the sequence number is taken after the write lock is released, right? Ning
