February 16, 201214 yr Hi guys, Another bug for you all to fix! If I have a TM backup going the same time as i'm streaming 1080p content via PS3 Media Server (no transcoding) it stutters, as it's not buffering, as it's not transcoding. This shouldn't happen because it's reading from disk 3/4 (Media share) and TM is directly to disk1 via no shares. So why is a TM backup, and the related IO interfering with a purely read process? Ideas?
February 16, 201214 yr Author Mac Pro 10.7.3 via Gigabit Ethernet. PS3 is also via Gigabit ethernet to the same NIC, via my Airport Extreme. I know the NIC can do >100MB/sec read/write so it's not a network b/w issue. Barely peaks at 20MB/sec usually as it's only backing up small files. Syslog shows nothing out of the ordinary, this is covering the period when the stuttering occurs: Feb 16 18:36:55 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:37:55 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:39:55 Tower last message repeated 2 times Feb 16 18:41:55 Tower last message repeated 2 times Feb 16 18:42:55 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:43:17 Tower kernel: mdcmd (23): clear (unRAID engine) Feb 16 18:43:55 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:44:55 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:46:55 Tower last message repeated 2 times Feb 16 18:47:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:48:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:49:37 Tower kernel: mdcmd (24): clear (unRAID engine) Feb 16 18:49:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:50:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 18:52:56 Tower last message repeated 2 times Feb 16 18:54:56 Tower last message repeated 2 times Feb 16 18:56:56 Tower last message repeated 2 times Feb 16 18:58:56 Tower last message repeated 2 times Feb 16 18:59:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:00:03 Tower in.telnetd[29406]: connect from 192.168.1.4 (192.168.1.4) (Routine) Feb 16 19:00:03 Tower login[29407]: invalid password for 'jamescullingham' on '/dev/pts/0' from 'MacBook-Pro' (Logins) Feb 16 19:00:04 Tower login[29407]: ROOT LOGIN on '/dev/pts/0' from 'MacBook-Pro' (Logins) Feb 16 19:00:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:01:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:03:56 Tower last message repeated 2 times Feb 16 19:05:56 Tower last message repeated 2 times Feb 16 19:07:56 Tower last message repeated 2 times Feb 16 19:09:56 Tower last message repeated 2 times Feb 16 19:10:46 Tower kernel: mdcmd (25): spindown 0 (Routine) Feb 16 19:10:47 Tower kernel: mdcmd (26): spindown 1 (Routine) Feb 16 19:10:48 Tower kernel: mdcmd (27): spindown 2 (Routine) Feb 16 19:10:48 Tower emhttp: shcmd (48): /usr/sbin/hdparm -y /dev/sda $stuff$> /dev/null (Drive related) Feb 16 19:10:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:11:56 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:13:57 Tower last message repeated 2 times Feb 16 19:15:57 Tower last message repeated 2 times Feb 16 19:17:57 Tower last message repeated 2 times Feb 16 19:19:57 Tower last message repeated 2 times Feb 16 19:21:57 Tower last message repeated 2 times Feb 16 19:23:57 Tower last message repeated 2 times Feb 16 19:25:57 Tower last message repeated 2 times Feb 16 19:27:57 Tower last message repeated 2 times Feb 16 19:29:57 Tower last message repeated 2 times Feb 16 19:30:08 Tower kernel: mdcmd (28): spindown 0 (Routine) Feb 16 19:30:08 Tower kernel: mdcmd (29): spindown 1 (Routine) Feb 16 19:30:09 Tower kernel: mdcmd (30): spindown 2 (Routine) Feb 16 19:30:09 Tower emhttp: shcmd (49): /usr/sbin/hdparm -y /dev/sda $stuff$> /dev/null (Drive related) Feb 16 19:30:57 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:31:57 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:33:57 Tower last message repeated 2 times Feb 16 19:34:30 Tower kernel: mdcmd (31): spindown 3 (Routine) Feb 16 19:34:31 Tower kernel: mdcmd (32): spindown 4 (Routine) Feb 16 19:45:58 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:47:00 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 19:49:00 Tower last message repeated 2 times Feb 16 19:51:00 Tower last message repeated 2 times Feb 16 19:53:00 Tower last message repeated 2 times Feb 16 19:55:00 Tower last message repeated 2 times Feb 16 19:57:00 Tower last message repeated 2 times Feb 16 19:59:00 Tower last message repeated 2 times Feb 16 20:01:00 Tower last message repeated 2 times Feb 16 20:01:27 Tower kernel: mdcmd (33): spindown 2 (Routine) Feb 16 20:01:27 Tower kernel: mdcmd (34): spindown 3 (Routine) Feb 16 20:01:28 Tower kernel: mdcmd (35): spindown 4 (Routine) Feb 16 20:01:28 Tower emhttp: shcmd (50): /usr/sbin/hdparm -y /dev/sda $stuff$> /dev/null (Drive related) Feb 16 20:02:00 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 20:03:00 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 20:05:00 Tower last message repeated 2 times Feb 16 20:07:01 Tower last message repeated 2 times Feb 16 20:09:01 Tower last message repeated 2 times Feb 16 20:11:01 Tower last message repeated 2 times Feb 16 20:12:01 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 20:12:50 Tower kernel: mdcmd (36): spindown 0 (Routine) Feb 16 20:12:50 Tower kernel: mdcmd (37): spindown 1 (Routine) Feb 16 20:46:02 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 20:47:02 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Feb 16 20:48:03 Tower logger: Thu Feb 16 18:17:54 GMT 2012 - Hard Drives active, resetting counter Full Syslog here: http://lime-technology.com/forum/index.php?topic=18363.msg164393#msg164393 (Ignore the Sil3132 error - it's not in the machine anymore)
February 16, 201214 yr I'll have to try it on my machine at home. Have a similar setup, I'll let you know how it behaves.
February 17, 201214 yr Author No. Device 1: Mac Pro doing IO to the server - Time Machine, or syncing photo files - nothing above 30MB/sec. Device 2: Playstation 3 being stream to from PS3 Media Server which is running on the machine itself. The issue is PS3 Media Server reading from the disks, as if I tell it to encode and cache 400MB, it plays until the buffer empties... So UnRAID can't do two IO ops to three different disks (Parity + disk 1 = TM) and read from disk 3.
February 17, 201214 yr I used to have a similar problem using and older Mac mini playing XMBC. XBMC would stutter once an hour, when TM was active. I started using CCC to do nightly backups, disabled TM on the Mini, and the problem when away. What CPU is in your server?
February 18, 201214 yr Author I used to have a similar problem using and older Mac mini playing XMBC. XBMC would stutter once an hour, when TM was active. I started using CCC to do nightly backups, disabled TM on the Mini, and the problem when away. What CPU is in your server? Quad 3.2GHz Phenom X4. CPU usage is negligible. <25% on all cores. For some reason, if parity is being accessed it all goes to pot
February 18, 201214 yr Author I wonder if a cache drive would make a difference. Maybe. BUt it's irrelevant. UnRAID should be able to manage completely separate IO. Ignoring network IO: If I am reading from disk 3 at 50MB/sec I should still be able to read and write to disk 0 and disk 1 at the max parity write speed of ~40MB/sec.
Archived
This topic is now archived and is closed to further replies.