IPL commit (tx:
Hi Y'all. There a message that occasionally pops up in the message log; here's one from December of last year: 12/29/25 21:56:19 Warning: IPL commit (tx:396465e28) is suspended, waiting for an open transaction from the oldest logical log (< some seq num >) to complete. When I googled the above it sent me to an HCL page with a similar message but gave advice useless to my situation. Here's how I understand the message but I believe my understanding is flawed. Suppose I have 1000 log files and the log sequence number of the oldest log file is 100001. I am currently at log sequence number 100123. There is a still a transaction open whose BEGIN record sits in log[100001]. It is still within the bounds of LTXHWM (or we would have gotten an LTX rollback message). But somehow, something triggered this IPL warning. There is obviously a difference between this IPL warning and the LTX rollback. We have procedures to seek out and kill the culprit session and certainly don't need to add a log (as per that aforementioned useless advice). So, after an uncharacteristically short explanation, here are my questions: What triggers this IPL warning, since it's not LTX? By my understanding, if the current log were 100002 - just in the next sequence number, there are likely MANY transactions with their BEGIN in log[100001]. Yet thatdid not trigger the IPL warning. What log sequence # would we have to be at to trigger this? I also understand that IPL is a TLA for I ndex P age L ogging. What in blazes does this situation with log files have to do with logging of index pages? Will someone please fill the gap between my ears on this issue?♀ Thanks much! ------------------------------ +-----------------------------------------------------------+ | I am pleased to report that I had no problems today. | | I had only issues, opportunities, challenges and valuable | | learning experiences. | +------------------------------------------ Jacob S --------+ ------------------------------
Jacob: You are in luck as I just explained this one to another client: As I explained to that client, the proximate cause is usually not having enough logical logs and applications that are holding transactions open for a relatively long time. Art Art S. Kagel, President and Principal Consultant ASK Database Management www.askdbmgt.com Blog: http://informix-myview.blogspot.com/ Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on the IIUG, nor any other organization with which I am associated either explicitly, implicitly, or by inference. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves.
Well, it's not a "change to an index page" that's being logged here, but rather an entire new index: with LOG_INDEX_BUILDS on, a new index, once fully built, is being copied to the logical log, page by page, so a secondary server would save building the index on its own while having a guaranteed binary copy of the primary's index. Depending on the size of such index, a large amount of logical log space can be consumed which in turn could compromise other parts of the system, esp. legitimate larger/longer other transactions suddenly running a risk of becoming LTX. To reduce this risk, IPL periodically checks for older transactions and temporarily suspends itself until such old guy is done. ------------------------------ Andreas Legner Informix Dev HCL Software ------------------------------