Re: [zfs-discuss] VMGuest IOMeter numbers
Hello Mark, I assume you have a read-intensive workload, not many synchronous writes, so leave out the ZIL, please try: * configure the controller to show individual disks, no RAID * create one large striped pool (zpool create tank c0t0d{1,2,3,4,5}) * if your SSD is c0t0d6, use it as an L2ARC (zpool create tank c0t0d{1,2,3,4,5} cache c0t0d6), not a ZIL * use the 134 developer release from here: http://www.genunix.org/distributions/indiana/ * give 80% of all the memory that your box has to the Opensolaris instance that is serving the ZFS I'm new here, so others may improve on these suggestions. Best regards, Oliver -- This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] NexentaStor Community edition 3.0.3 released
Hello, this may not apply to your machine. I have two changes to your setup: * Opensolaris instead of Nexenta * DL585G1 instead of your DL380G4 Here's my problem: reproducible crash after a certain time (1:30h in my case). Explanation: the HP machine has enterprise features (ECC RAM) and performs scrubbing of the RAM, just as you could scrub ZFS disks; with the 4 AMD dual core CPUs, the memory is divided into 4 chunks and when the scrubber hits a hole, then the machine crashes without so much as a crashdump Solution: add the following to /etc/system set snooping=1 set pcplusmp:apic_panic_on_nmi=1 set cpu_ms.AuthenticAMD.15:ao_scrub_policy = 1 set cpu_ms.AuthenticAMD.15:ao_scrub_rate_dcache = 0 set cpu_ms.AuthenticAMD.15:ao_scrub_rate_l2cache = 0 set mc-amd:mc_no_attach=1 set disable_memscrub = 1 Best regards, Oliver -- This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] /bin/cp vs /usr/gnu/bin/pc
Hello, I came across this blog post: http://kevinclosson.wordpress.com/2007/03/15/copying-files-on-solaris-slow-or-fast-its-your-choice/ and would like to hear from you performance gurus how this 2007 article relates to the 2010 ZFS implementation? What should I use and why? Thanks, Oliver -- This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] mirror resilver @500k/s
Hello Everybody, thank you for your support. I have been able to find a sustained 50-70mb resilvering with the "iostat -x 10" command. On one out of 3 discs. The other two discs are now on their way back to the vendor and I hope to be able to report better success when I get them back. Thanks again, Oliver -- This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
Re: [zfs-discuss] mirror resilver @500k/s
Hello Will, thank you for the explanation of "zpool iostat -v data" without any further arguments! I will run the two suggested commands when I get back from work. Yes, the 20gb have taken about 12h to resilver. Now there's just 204gb left to do ... Thanks to everyone for your replies, Oliver (now back to my temporarily disabled user ID) -- This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
[zfs-discuss] mirror resilver @500k/s
Hello, I'm a grown-up and willing to read, but I can't find where to read. Please point me to the place that explains how I can diagnose this situation: adding a mirror to a disk fills the mirror with an apparent rate of 500k per second. 1) what diagnostic information should I look at (and perhaps provide to you people here)? 2) how should I have gone about seeking help for a problem like this? 3) on a related note -- why is "zpool status -v data" slower to run as root than it is as a normal user? Thanks for your time! Oliver os10...@giant:~$ (zpool status -v data; zpool iostat -v data; dmesg | tail -5) | egrep -v '^$' pool: data state: ONLINE status: One or more devices is currently being resilvered. The pool will continue to function, possibly in a degraded state. action: Wait for the resilver to complete. scrub: resilver in progress for 12h13m, 45.74% done, 14h29m to go config: NAMESTATE READ WRITE CKSUM dataONLINE 0 0 0 mirror-0 ONLINE 0 0 0 c9t3d0 ONLINE 0 0 0 c9t1d0 ONLINE 0 0 0 mirror-1 ONLINE 0 0 0 c9t2d0 ONLINE 0 0 0 c9t0d0 ONLINE 0 0 0 20.4G resilvered errors: No known data errors capacity operationsbandwidth poolalloc free read write read write -- - - - - - - data 755G 1.76T 67 3 524K 14.1K mirror 530G 1.29T 6 1 40.8K 5.89K c9t3d0 - - 3 1 183K 6.04K c9t1d0 - - 3 0 183K 6.04K mirror 224G 472G 60 1 484K 8.24K c9t2d0 - - 13 0 570K 4.05K c9t0d0 - - 0 34 17 490K -- - - - - - - May 13 10:33:38 giant genunix: [ID 936769 kern.notice] sv0 is /pseudo/s...@0 May 13 10:33:38 giant pseudo: [ID 129642 kern.notice] pseudo-device: ii0 May 13 10:33:38 giant genunix: [ID 936769 kern.notice] ii0 is /pseudo/i...@0 May 13 10:34:34 giant su: [ID 810491 auth.crit] 'su root' failed for os1 on /dev/pts/4 May 13 20:44:09 giant pcplusmp: [ID 805372 kern.info] pcplusmp: ide (ata) instance 1 irq 0xf vector 0x45 ioapic 0x4 intin 0xf is bound to cpu 6 -- This message posted from opensolaris.org ___ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss