updatedb/locate difficulty

UB_KMA · 12-28-2002, 04:57 AM

While messing around trying to pick a new linux kernel (I finally decided on a custom 2.4.19 from Lunar with a lot of processor specific tweaks added) I got a segmentation fault during an updatedb command...at that point I was actually running the 2.4.20-lq. After that, nomatter which kernel I was running, updatedb seems to run ok, but I get a segmetation fault while running a subsequent locate command, and the results of the locate are truncated to just a few entries when there might be dozens of results. Everything else seems to be running flawlessly, and lightning fast with the final kernel I picked, but I needs me locate, and this is beyond my ken!

Re-install bash? I may try that...any suggestions or exlanations are appriciated!

jetblackz · 12-28-2002, 12:26 PM

You might want to delete

/usr/var/locatedb

or

/var/lib/locate/locatedb

and recreate it using updatedb

UB_KMA · 12-28-2002, 02:05 PM

Well, it sounded like a good idea! Deleted updatedb (it was actually in /var/cache), and reran updatedb. It recreated the file, so I did a locate...which again terminated in a segmentation fault

.

jetblackz · 12-29-2002, 12:16 PM

1) not enough swap space
2) bad memory
3) run e2fsck
4) read http://www.bitwizard.nl/sig11/

UB_KMA · 12-29-2002, 07:28 PM

Very interesting info in that link..I knew what segmentation fault was (after looking it up) but had no idea as to the many possible causes. This only happens during a locate command, so I am suspecting that it is actually trying to access unauthorized memory space and it is not just a bug or error. I know that my profiles says I am running slack, but I am currently running a source based setup made from scratch which is using /devfs. Now, I have /dev/proc mounted, and many other memory only resident filesystems that are possibly being scanned during updatedb. Since these mounted filesystems are in memory only, is it possible a seg fault could result during a locate?
I have tested doing a limited updatedb, specifying to only catalog certain directories (after deleting the old updatedb file), and a subsequent locate command doesn't result in the fault and completes sucessfully.

To respond to your suggestions:
I have a rediculous amount of swap space (large).
I really hope it isn't my memory!
I will run e2fsk and see what results...