-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Pawel, I don't know if this is going to be helpful, but I did the following experiment : I've disabled all tuning in loader.conf, all values for arc_max, kmem_size and kmem_size_max are the defaults computed for amd64 machine with 4G of RAM. I've put debug printf()'s in the loop in vm/vm_kern.c:kern_malloc() printing the number of loops. I've increased the loop count in kern_malloc() from 8 to 100. I've also put debug printf()'s in the arc_lowmem() function in zfs's ARC cache, this should be the function that is called when the "vm_lowmem" event fires. Then I started one bonnie++ process and watched /var/log/messages. Several times the loopcount in kern_malloc() jumped from 0 to 2 or 3 maximum, and everything continued to work, but after 20-30 minutes they started to increase steadily and it took the machine another maybe 20 minutes to reach 99 loops and then panic again with kmem_map too small. I hope this makes sense. - -- Regards, Nikolay Denev -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (Darwin) iEYEARECAAYFAkkugEcACgkQHNAJ/fLbfrnJGACfSHq/dKJFBBecMWfElI35m/2C mwQAoLsafDZI/lALZO/Xr9i/MYbdkiUm =QTnU -----END PGP SIGNATURE-----Received on Thu Nov 27 2008 - 10:11:09 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:38 UTC