Hello Tomas, Wednesday, January 3, 2007, 10:32:39 AM, you wrote:
TÖ> Hello. TÖ> Having some hangs on a snv53 machine which is quite probably ZFS+NFS TÖ> related, since that's all the machine do ;) TÖ> The machine is a 2x750MHz Blade1000 with 2GB ram, using a SysKonnect TÖ> 9821 GigE card (with their 8.19.1.3 skge driver) and two HP branded MPT TÖ> SCSI cards. Normal load is pretty much "read all you can" with misc TÖ> tarballs and isos since it's an NFS backend to our caching http/ftp TÖ> cluster delivering free software to the world. TÖ> What happens is that the machine just stops responding.. it can respond TÖ> to ping for a while (while userland is non-responsive, including TÖ> console) but after a while, that stops too.. TÖ> Produced a panic to get a dump and tried ::memstat; TÖ> unterweser:/scratch/070103# mdb unix.0 vmcore.0 TÖ> Loading modules: [ unix krtld genunix specfs dtrace ufs scsi_vhci pcisch TÖ> ssd fcp fctl qlc md ip hook neti sctp arp usba s1394 nca lofs zfs random TÖ> sd nfs ptm cpc ] >> ::memstat TÖ> Page Summary Pages MB %Tot TÖ> ------------ ---------------- ---------------- ---- TÖ> Kernel 250919 1960 98% TÖ> Anon 888 6 0% TÖ> Exec and libs 247 1 0% TÖ> Page cache 38 0 0% TÖ> Free (cachelist) 405 3 0% TÖ> Free (freelist) 4370 34 2% TÖ> Total 256867 2006 TÖ> Physical 253028 1976 TÖ> That doesn't seem too healthy to me.. probably something kernely eating TÖ> up everything and the machine is just swapping to death or something.. TÖ> A dump from live kernel with mdb -k after 1.5h uptime; TÖ> Page Summary Pages MB %Tot TÖ> ------------ ---------------- ---------------- ---- TÖ> Kernel 212310 1658 83% TÖ> Anon 11307 88 4% TÖ> Exec and libs 2418 18 1% TÖ> Page cache 18400 143 7% TÖ> Free (cachelist) 4383 34 2% TÖ> Free (freelist) 8049 62 3% TÖ> The tweaks I have are: TÖ> set ncsize = 500000 TÖ> set nfs:nrnode = 50 TÖ> set zfs:zil_disable=1 TÖ> set zfs:zfs_vdev_cache_bshift=14 TÖ> set zfs:zfs_vdev_cache_size=0 Comment out ncsize and reboot - should help. I had similar behavior with increased ncsize from default on ZFS+NFS. Mark suggested to comment it out and it just works since then. I was also told increasing ncsize with ZFS isn't needed that much (if at all) - however I haven't done any testing and I don't know technical details how DNLC is used with ZFS to be sure. -- Best regards, Robert mailto:[EMAIL PROTECTED] http://milek.blogspot.com _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
