Today you sent me mail regarding Exiting of programs during an rrd_update: *> I have one more thing that I'd like thought about before the *> final release of RRDtool. *> *> What happens to an RRD file if the program running rrd_update *> is control-c during an rrd_update. Can the file be corrupted *> or screwed up in any way? If so, I'd like to see rrd_update *> potentially open writing to a temporary file, create the *> file, close the file, and then do a rename to the final *> name. In this way if anything goes wrong before the *> rename, no data is lost or screwed up.
Hi Blair, the tmp file aprocah is not a solution at this would defeat the whole point of round robin archives where the file is EDITED insted of rewritten ... We could put a signal handler in there which made rrd_update and rrd_create ignore all signals when it executes the fwrites ... cheers tobi *> *> Comments, *> Blair -- ______ __ _ /_ __/_ / / (_) Oetiker, hacking RRDTOOL&MRTG @ Caida for the Summer / // _ \/ _ \/ / +1 619 822 0882 / [EMAIL PROTECTED] / www.caida.org/~tobi /_/ \.__/_.__/_/ _______________________________________________________ -- * To unsubscribe from the rrd-developers mailing list, send a message with the subject: unsubscribe to [EMAIL PROTECTED]
