The planner has an "ERROR" to make the estimate, but than later the dump itself FAILs as well. So no backup is made of that particular DLE.
Regards, Charles On Tue, 11 Dec 2012 09:04:46 +0300 Alan Orth <[email protected]> wrote: > Hi, All. > > It's good that you brought this up on the mailing list, I was just > about to ask! I've been having problems with estimation timeouts > lately too, so I'll try some of these tips to fix it. > > What confused me initially was why estimation failures are classified > as "FAIL"? It's quite worrying when you wake up in the morning to > find last night's backups have FAILED. Shouldn't the classification > be more similar to something like the STRANGE errors (where files > have changed during backup, for example)? > > Cheers, > > Alan > > On 12/10/2012 06:05 PM, Charles Stroom wrote: > > Hi, the forwarded email below was meant to go to the list, but I > > noticed later it was only to 1 recepient. Hence the forward. > > > > Regards, Charles > > > > > > > > Begin forwarded message: > > > > Date: Sat, 8 Dec 2012 22:08:48 +0100 > > From: Charles Stroom <[email protected]> > > To: Jens Berg <[email protected]> > > Subject: Re: failure "estimate of level x timed out" > > > > > > So far, so good. Since I have increased etimeout to 14400 AND set > > "estimate calcsize" I have had 2 backups without failures. The only > > thing I don't know yet which parameter did the trick. I keep my > > fingers crossed. > > > > Thanks both of you. > > > > Charles > > > > > > > > On Thu, 06 Dec 2012 09:49:28 +0100 > > Jens Berg <[email protected]> wrote: > > > >> I would suggest to increase etimeout to a much bigger value, let's > >> say 14400 or so and see if the estimates finish at all then. If > >> they still fail, I would take a closer look on the health of the > >> hard discs... Another option could be to change the estimate > >> method for the dump type you are using, e.g. if you are using > >> "dumptype user-tar" for the DLEs, put an "estimate calcsize" in > >> the definition of "dumptype user-tar". The results of that > >> estimate method will be less accurate than the ones from the > >> default method but it executes faster. > >> > >> Best > >> Jens > >> > > -- > Alan Orth > [email protected] > http://alaninkenya.org > http://mjanja.co.ke > "I have always wished for my computer to be as easy to use as my > telephone; my wish has come true because I can no longer figure out > how to use my telephone." -Bjarne Stroustrup, inventor of C++ > -- Charles Stroom email: charles at no-spam.stremen.xs4all.nl (remove the "no-spam.")
