[Progress Communities] [Progress OpenEdge ABL] Forum Post: RE: Buffer locked (or not locked?) on target database

  • Thread starter Thread starter Dapeng Wu
  • Start date Start date
Status
Not open for further replies.
D

Dapeng Wu

Guest
One suspect of this long pause is the replication agent, although I can't explain why Promon didn't show this. If that is indeed the case, say for some reason, "buffer locks" for agent is missed by Promon, what could happen is there was an index split operation, and only part of the operation was replicated to the target while the clients were reading the index. These clients will have to wait till the whole index split is completed before they can get the root block. This is done to protect the reads because we don't want that happen in the middle of a split. In the meantime, for the index split operation on the source, even though it can finish pretty quick, the AI notes for this operation may take a while to fully replicate to the target (network latency, for example). And only after that, the index blocks will be released by the agent. Again, this theory is based on the assumption that the agent was holding a X lock on the root block, which for some reason, we didn't see in this case.

Continue reading...
 
Status
Not open for further replies.
Back
Top