Kaygee Posted November 17, 2009 Share Posted November 17, 2009 It does not look like a shared IRQ issue. As far as IRQ's being shuffled, anything above 15 (I believe) is programmed in software Although I don't know what PCI-MSI-edge is. PCI-MSI is an attempt in PCI 2.3 and 3.0 devices to remove the requirement for hardware interupts altogether. MSI = Message Signalled Interrupt. In order to reduce pin count the document claims (however all CPU manufacturers seem to be increasing pin counts not reducing them (Intel are upto socket 1366 vs amd socket 940). Chuckle I did. PCI-e uses INTx emulation and all INTx capable devices should be able to share interupts, therfore all PCI-e devices should be capable of sharing Interupts. MSI is a message based interupt service using a memory write to the APIC. MSI-X (extensions) allows 32 vectors per device to be requested. This looks to be a mechanism to allow MSI based IRQs to be serviced/directed to specific CPUs. Although I'm not 100% sure I'm reading that right... -edge means the apic is using edge triggering rather than level. Basic MSI how-to is here: http://lwn.net/Articles/44139/ or http://devresources.linux-foundation.org/dev/robustmutexes/src/fusyn.hg/Documentation/MSI-HOWTO.txt Grabbing a cat /proc/interrupts when the system doesn't autneg 1000Mb/s would be very interesting, also if the link drops to 100Mb/s. It would indicate an issue with MSI implementation in the driver or a lack of available MSI resources. The device should then switch to INTx behaviour, it will be interesting to see what INT it drops too. Quote Link to comment
limetech Posted November 17, 2009 Author Share Posted November 17, 2009 Well, this makes a lot of sense Tom. It also happens to me on my 6 onboard SATA controllers (I don't use any expansion cards atm). Makes a heck of a lot of sense though, I can reproduce the issue. What combination of the 6 on-board ports are 'paired'? Actually if you could select/copy/paste the relevant lines on the Devices page that would be helpful. Quote Link to comment
limetech Posted November 17, 2009 Author Share Posted November 17, 2009 Are users of SATA port multipliers affected by this as well? Probably not, but that is a good question. I have a port multiplier and this issue is absolutely killing me. Silicon Image 3726 with a Sil 3132-based card. Please provide more details of your config and select/copy/paste the relevant lines of the Devices page - I'm curious if the set of drives on one controller is > 2 (probably is for command-based port multipliers). Quote Link to comment
Kaygee Posted November 17, 2009 Share Posted November 17, 2009 @Brainbone What MB are you using? What mode are your SATA ports using (if configurable in the bios - IDE, Enhanced IDE, Legacy IDE, RAID or AHCI). Might be usefull to post a copy of you device setting like the ones below: parity device: pci-0000:00:09.0-scsi-3:0:0:0 (sdf) ata-SAMSUNG_HD103UJ_x disk1 device: pci-0000:00:09.0-scsi-2:0:0:0 (sde) ata-SAMSUNG_HD103UJ_x disk2 device: pci-0000:00:09.0-scsi-0:0:0:0 (sdc) ata-SAMSUNG_HD103UJ_x disk3 device: pci-0000:00:09.0-scsi-1:0:0:0 (sdd) ata-SAMSUNG_HD103UJ_x disk4 device: pci-0000:00:09.0-scsi-4:0:0:0 (sdg) ata-ST3500630AS_x disk5 device: pci-0000:00:09.0-scsi-5:0:0:0 (sdh) ata-ST3500630AS_x disk6 device: pci-0000:01:07.0-scsi-2:0:0:0 (sdb) ata-ST3500630AS_x Quote Link to comment
limetech Posted November 17, 2009 Author Share Posted November 17, 2009 @Brainbone What MB are you using? What mode are your SATA ports using (if configurable in the bios - IDE, Enhanced IDE, Legacy IDE, RAID or AHCI). Might be usefull to post a copy of you device setting like the ones below: parity device: pci-0000:00:09.0-scsi-3:0:0:0 (sdf) ata-SAMSUNG_HD103UJ_x disk1 device: pci-0000:00:09.0-scsi-2:0:0:0 (sde) ata-SAMSUNG_HD103UJ_x disk2 device: pci-0000:00:09.0-scsi-0:0:0:0 (sdc) ata-SAMSUNG_HD103UJ_x disk3 device: pci-0000:00:09.0-scsi-1:0:0:0 (sdd) ata-SAMSUNG_HD103UJ_x disk4 device: pci-0000:00:09.0-scsi-4:0:0:0 (sdg) ata-ST3500630AS_x disk5 device: pci-0000:00:09.0-scsi-5:0:0:0 (sdh) ata-ST3500630AS_x disk6 device: pci-0000:01:07.0-scsi-2:0:0:0 (sdb) ata-ST3500630AS_x Yes, for everyone following this thread, it would be useful to see your device settings as Kaygee has posted here, and whether or not your system exhibits the 'stalled media stream' behavior. @Kaygee - from the looks of your devices info, you system does not exhibit the stalled media behavior, correct? Quote Link to comment
Stokkes Posted November 17, 2009 Share Posted November 17, 2009 What combination of the 6 on-board ports are 'paired'? Actually if you could select/copy/paste the relevant lines on the Devices page that would be helpful. Here it is.. parity device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdb) ata-ST31500341AS_9VS0NMBJ disk1 device: pci-0000:00:1f.5-scsi-1:0:0:0 (sdf) ata-ST31500341AS_9VS0BM7F disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sdd) ata-ST31500341AS_9VS0MCCM disk3 device: pci-0000:00:1f.2-scsi-0:0:1:0 (sdc) ata-ST31500341AS_9VS0BGPA cache device: pci-0000:00:1f.2-scsi-1:0:1:0 (sde) ata-WDC_WD5000AAKS-00YGA0_WD-WCAS82838546 Quote Link to comment
brainbone Posted November 17, 2009 Share Posted November 17, 2009 @Brainbone What MB are you using? What mode are your SATA ports using (if configurable in the bios - IDE, Enhanced IDE, Legacy IDE, RAID or AHCI). I'd have to double check the MB and the settings in BIOS, but my configuration until I get around to that: parity device: pci-0000:00:1f.5-scsi-1:0:0:0 (sdh) ata-Hitachi_HDT721010SLA360_STF607MH34HUTK disk1 device: pci-0000:00:1f.2-scsi-0:0:1:0 (sdd) ata-ST3500630AS_5QG2X1TT disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sde) ata-ST3500630AS_5QG2WW8Z disk3 device: pci-0000:00:1f.2-scsi-1:0:1:0 (sdf) ata-Hitachi_HDP725032GLA360_GEAR34RF1VTGTG disk4 device: pci-0000:00:1f.5-scsi-0:0:0:0 (sdg) ata-Hitachi_HDP725032GLA360_GEAR34RF1VSYMG disk5 device: pci-0000:03:00.0-scsi-0:0:0:0 (sda) ata-Hitachi_HDT721010SLA360_STF607MH34AXVK disk6 device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdc) ata-ST3500630AS_5QG2YGPK I do have the spin-up / stream delay on already spun-up issue (is there a short way of describing this issue?) EDIT: MB is DFI INFINITY P965-S Quote Link to comment
limetech Posted November 17, 2009 Author Share Posted November 17, 2009 What combination of the 6 on-board ports are 'paired'? Actually if you could select/copy/paste the relevant lines on the Devices page that would be helpful. Here it is.. parity device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdb) ata-ST31500341AS_9VS0NMBJ disk1 device: pci-0000:00:1f.5-scsi-1:0:0:0 (sdf) ata-ST31500341AS_9VS0BM7F disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sdd) ata-ST31500341AS_9VS0MCCM disk3 device: pci-0000:00:1f.2-scsi-0:0:1:0 (sdc) ata-ST31500341AS_9VS0BGPA cache device: pci-0000:00:1f.2-scsi-1:0:1:0 (sde) ata-WDC_WD5000AAKS-00YGA0_WD-WCAS82838546 Looks like your pairs are parity/disk3 and disk1/cache. What motherboard is this? Quote Link to comment
mcs Posted November 17, 2009 Share Posted November 17, 2009 What combination of the 6 on-board ports are 'paired'? Actually if you could select/copy/paste the relevant lines on the Devices page that would be helpful. Here it is.. parity device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdb) ata-ST31500341AS_9VS0NMBJ disk1 device: pci-0000:00:1f.5-scsi-1:0:0:0 (sdf) ata-ST31500341AS_9VS0BM7F disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sdd) ata-ST31500341AS_9VS0MCCM disk3 device: pci-0000:00:1f.2-scsi-0:0:1:0 (sdc) ata-ST31500341AS_9VS0BGPA cache device: pci-0000:00:1f.2-scsi-1:0:1:0 (sde) ata-WDC_WD5000AAKS-00YGA0_WD-WCAS82838546 Looks like your pairs are parity/disk3 and disk1/cache. What motherboard is this? Should this be parity/disk3 and disk2/cache Quote Link to comment
prostuff1 Posted November 17, 2009 Share Posted November 17, 2009 Well, this makes a lot of sense Tom. It also happens to me on my 6 onboard SATA controllers (I don't use any expansion cards atm). Makes a heck of a lot of sense though, I can reproduce the issue. What combination of the 6 on-board ports are 'paired'? Actually if you could select/copy/paste the relevant lines on the Devices page that would be helpful. For reference and since you requested it here is the devices page as mine currently stands parity device: pci-0000:00:1f.2-scsi-3:0:0:0 (sde) ata-ST31500341AS_9VS1392C disk1 device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdb) ata-ST3750640AS_5QD5ELLF disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sdc) ata-WDC_WD5000AAKS-00TMA0_WD-WCAPW2595673 disk3 device: pci-0000:00:1f.2-scsi-4:0:0:0 (sdf) ata-WDC_WD5000AAKS-00TMA0_WD-WCAPW2132942 disk4 device: pci-0000:05:00.0-scsi-0:0:0:0 (sdh) ata-SAMSUNG_HD753LJ_S13UJ1MQ330294 disk5 device: pci-0000:05:00.0-scsi-1:0:0:0 (sdi) ata-ST31000333AS_6TE0DYD1 disk6 device: pci-0000:00:1f.2-scsi-5:0:0:0 (sdg) ata-ST31500341AS_6VS04GWN cache device: pci-0000:00:1f.2-scsi-2:0:0:0 (sdd) ata-WDC_WD1600JD-40HBC0_WD-WCAL94036151 This is on an Abit AB9 Pro on all the internal ports. I am using all but one of my onboard SATA ports which is labeled SATA7 in my motherboard manual. If there is anything else I can provide let me know and I will be happy to oblige. I have noticed the spin up problem but frankly it is not that big of a concern to me. So my movie pauses for a second or two, I can deal with it as it happens very rarely for me. Quote Link to comment
WeeboTech Posted November 17, 2009 Share Posted November 17, 2009 Some background: most of you will recall when IDE drives ruled the PC world and each IDE "controller" could have two hard drives attached, a master and a slave. What is less common knowledge is that, at any one time, you could execute an I/O request on either the master drive or the slave drive, but not both simultaneously. This is a low-level "detail" which, for a variety of reasons, doesn't normally have much impact on the applications using the hard drives. Now SATA rules the PC world, but many early SATA controllers are simply IDE controllers in disguise. For example, the Promise SATA300 TX4 controller has 4 SATA ports, but internally, there are actually two SATA controllers, each controlling two SATA ports. This h/w has the same limitation: if you execute an I/O operation on one SATA port, and then try to execute another SATA operation on the other port of the same internal controller, that second I/O operation can not be executed until the first one finishes. Tom, thank you for the explanation. I always knew the issue existed with IDE controllers, and theorized that it existed with SATA, but now you have solidified my understanding of it. This is exactly the reason I use SCSI or SAS for my heavily used workstations and servers. I wonder how this affects people with Parity drives that share a controller with a data drive. I always recommend using only 1 drive per IDE cable. Should there be a recommendation to suggest insuring parity is not sharing a controller. @Kaygee, thanks for the explanation. Tom: Perhaps you can make the "pair-spin-up" a tunable feature, to satisfy both those who want their movie watching experience to not suffer, and to please those who care about every "watt" consumed and could care less about the spin-up-delay pause of their data stream. Joe L. I would prefer the pause, stutter or getting new hardware vs spinning up a drive I don't have to. A a tunable would be really nice. Quote Link to comment
Joe L. Posted November 17, 2009 Share Posted November 17, 2009 Yes, for everyone following this thread, it would be useful to see your device settings as Kaygee has posted here, and whether or not your system exhibits the 'stalled media stream' behavior. I do experience this "stalled media stream" behavior. My MB is the original Intel board you shipped. /dev/sdf parity=pci-0000:00:1f.2-scsi-1:0:0:0 /dev/hdb disk1=pci-0000:00:1f.1-ide-0:1 /dev/hdc disk2=pci-0000:00:1f.1-ide-1:0 /dev/hdd disk3=pci-0000:00:1f.1-ide-1:1 /dev/hde disk4=pci-0000:02:00.0-ide-0:0 /dev/hdf disk5=pci-0000:02:00.0-ide-0:1 /dev/hdg disk6=pci-0000:02:00.0-ide-1:0 /dev/hdi disk7=pci-0000:02:02.0-ide-1:1 /dev/hdj disk8=pci-0000:02:02.0-ide-1:0 /dev/hda disk10=pci-0000:00:1f.1-ide-0:0 /dev/sde disk11=pci-0000:00:1f.2-scsi-0:0:0:0 Being mostly IDE controller based, I'd expect it to act as you described. Joe L. Quote Link to comment
Kaygee Posted November 17, 2009 Share Posted November 17, 2009 @Kaygee - from the looks of your devices info, you system does not exhibit the stalled media behavior, correct? You are correct sir. I tried and tried to repro but couldn't, I tried both SATA mode, AHCI mode in the bios and a couple of drives hooked to the SIL3124, I will try my PMP hooked to the SIL3124 tomorrow to see if that exhibits the behavior (it shouldnt as SIL3124 definetly uses FIS based switching). Its too late to start dismantling the server tonight. Quote Link to comment
limetech Posted November 17, 2009 Author Share Posted November 17, 2009 What combination of the 6 on-board ports are 'paired'? Actually if you could select/copy/paste the relevant lines on the Devices page that would be helpful. Here it is.. parity device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdb) ata-ST31500341AS_9VS0NMBJ disk1 device: pci-0000:00:1f.5-scsi-1:0:0:0 (sdf) ata-ST31500341AS_9VS0BM7F disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sdd) ata-ST31500341AS_9VS0MCCM disk3 device: pci-0000:00:1f.2-scsi-0:0:1:0 (sdc) ata-ST31500341AS_9VS0BGPA cache device: pci-0000:00:1f.2-scsi-1:0:1:0 (sde) ata-WDC_WD5000AAKS-00YGA0_WD-WCAS82838546 Looks like your pairs are parity/disk3 and disk1/cache. What motherboard is this? Should this be parity/disk3 and disk2/cache Yes you are correct. Quote Link to comment
BRiT Posted November 17, 2009 Share Posted November 17, 2009 Might some of the stuttering also be related to the BKL used by ReiserFS? FWIW, this is my SATA tables when BIOS is set to AHCI. I haven't noticed any stutter, but for now all my media is on one drive. Nov 17 18:46:39 Reaver emhttp: pci-0000:00:1f.2-scsi-0:0:0:0 (sda) ata-WDC_WD1001FALS-00J7B0_WD-WMATV1120303 Nov 17 18:46:39 Reaver emhttp: pci-0000:00:1f.2-scsi-1:0:0:0 (sdb) ata-WDC_WD20EADS-00R6B0_WD-WCAVY0211284 Nov 17 18:46:39 Reaver emhttp: pci-0000:00:1f.2-scsi-4:0:0:0 (sdc) ata-WDC_WD20EADS-00R6B0_WD-WCAVY0247937 Nov 17 18:46:39 Reaver emhttp: pci-0000:00:1f.2-scsi-5:0:0:0 (sdd) ata-WDC_WD20EADS-00R6B0_WD-WCAVY0252670 Quote Link to comment
Kaygee Posted November 18, 2009 Share Posted November 18, 2009 @Brit looks fine. @Brainbone parity device: pci-0000:00:1f.5-scsi-1:0:0:0 (sdh) ata-Hitachi_HDT721010SLA360_STF607MH34HUTK disk1 device: pci-0000:00:1f.2-scsi-0:0:1:0 (sdd) ata-ST3500630AS_5QG2X1TT disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sde) ata-ST3500630AS_5QG2WW8Z disk3 device: pci-0000:00:1f.2-scsi-1:0:1:0 (sdf) ata-Hitachi_HDP725032GLA360_GEAR34RF1VTGTG disk4 device: pci-0000:00:1f.5-scsi-0:0:0:0 (sdg) ata-Hitachi_HDP725032GLA360_GEAR34RF1VSYMG disk5 device: pci-0000:03:00.0-scsi-0:0:0:0 (sda) ata-Hitachi_HDT721010SLA360_STF607MH34AXVK disk6 device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdc) ata-ST3500630AS_5QG2YGPK Disk2 and 3 share. Disk 1 and 6 share. @Prostuff No shared/paired devices are listed... Quote Link to comment
lewcass Posted November 18, 2009 Share Posted November 18, 2009 It does not look like a shared IRQ issue. As far as IRQ's being shuffled, anything above 15 (I believe) is programmed in software Although I don't know what PCI-MSI-edge is. PCI-MSI is an attempt in PCI 2.3 and 3.0 devices to remove the requirement for hardware interupts altogether. MSI = Message Signalled Interrupt. In order to reduce pin count the document claims (however all CPU manufacturers seem to be increasing pin counts not reducing them (Intel are upto socket 1366 vs amd socket 940). Chuckle I did. PCI-e uses INTx emulation and all INTx capable devices should be able to share interupts, therfore all PCI-e devices should be capable of sharing Interupts. MSI is a message based interupt service using a memory write to the APIC. MSI-X (extensions) allows 32 vectors per device to be requested. This looks to be a mechanism to allow MSI based IRQs to be serviced/directed to specific CPUs. Although I'm not 100% sure I'm reading that right... -edge means the apic is using edge triggering rather than level. Basic MSI how-to is here: http://lwn.net/Articles/44139/ or http://devresources.linux-foundation.org/dev/robustmutexes/src/fusyn.hg/Documentation/MSI-HOWTO.txt Grabbing a cat /proc/interrupts when the system doesn't autneg 1000Mb/s would be very interesting, also if the link drops to 100Mb/s. It would indicate an issue with MSI implementation in the driver or a lack of available MSI resources. The device should then switch to INTx behaviour, it will be interesting to see what INT it drops too. Thanks. I will definitely check the interrupts again the next time I see a problem. So far it has been reliably reconnecting at 1000 with every WOL. This is consistent with past behavior. i.e., When it works at start-up, it continues to work. Quote Link to comment
brainbone Posted November 18, 2009 Share Posted November 18, 2009 @Brainbone What MB are you using? What mode are your SATA ports using (if configurable in the bios - IDE, Enhanced IDE, Legacy IDE, RAID or AHCI). I'd have to double check the MB and the settings in BIOS, but my configuration until I get around to that: parity device: pci-0000:00:1f.5-scsi-1:0:0:0 (sdh) ata-Hitachi_HDT721010SLA360_STF607MH34HUTK disk1 device: pci-0000:00:1f.2-scsi-0:0:1:0 (sdd) ata-ST3500630AS_5QG2X1TT disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sde) ata-ST3500630AS_5QG2WW8Z disk3 device: pci-0000:00:1f.2-scsi-1:0:1:0 (sdf) ata-Hitachi_HDP725032GLA360_GEAR34RF1VTGTG disk4 device: pci-0000:00:1f.5-scsi-0:0:0:0 (sdg) ata-Hitachi_HDP725032GLA360_GEAR34RF1VSYMG disk5 device: pci-0000:03:00.0-scsi-0:0:0:0 (sda) ata-Hitachi_HDT721010SLA360_STF607MH34AXVK disk6 device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdc) ata-ST3500630AS_5QG2YGPK I do have the spin-up / stream delay on already spun-up issue (is there a short way of describing this issue?) EDIT: MB is DFI INFINITY P965-S Checked BIOS and sure enough my SATA controller was set to IDE. Changed it to AHCI, and now have: parity device: pci-0000:00:1f.2-scsi-5:0:0:0 (sdh) ata-Hitachi_HDT721010SLA360_STF607MH34HUTK disk1 device: pci-0000:00:1f.2-scsi-2:0:0:0 (sde) ata-ST3500630AS_5QG2X1TT disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sdd) ata-ST3500630AS_5QG2WW8Z disk3 device: pci-0000:00:1f.2-scsi-3:0:0:0 (sdf) ata-Hitachi_HDP725032GLA360_GEAR34RF1VTGTG disk4 device: pci-0000:00:1f.2-scsi-4:0:0:0 (sdg) ata-Hitachi_HDP725032GLA360_GEAR34RF1VSYMG disk5 device: pci-0000:03:00.0-scsi-0:0:0:0 (sda) ata-Hitachi_HDT721010SLA360_STF607MH34AXVK disk6 device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdc) ata-ST3500630AS_5QG2YGPK Initial testing shows that the issue is gone. Quote Link to comment
limetech Posted November 18, 2009 Author Share Posted November 18, 2009 Might some of the stuttering also be related to the BKL used by ReiserFS? Always wondered that myself. There is a significant patch in the works to eliminate the BKL in reiserfs, discussed here. This patch is not in 2.6.31.x series & I haven't looked at 2.6.32 yet to see if it's in there [care to investigate this for me?]. FYI, the term "BKL" means "Big Kernel Lock". Originally with linux, when a driver or file system needed to do certain atomic operations it would acquire the BKL, do it's thing, then release the BKL. During the time BKL is held, kernel can do absolutely nothing else. There's been a longstanding effort in kernel development to get rid of the BKL because (I think) of it's impact in multi-core systems. Since the bit rates for media are relatively slow & there exists plenty of other potential bottlenecks, I don't think this is an issue whatsoever in unRAID. But interested readers can google "linux bkl" and find lots of discussion on the subject. Quote Link to comment
Guzzi Posted November 18, 2009 Share Posted November 18, 2009 Are users of SATA port multipliers affected by this as well? Probably not, but that is a good question. in theory, when using FIS-based switching, you should not be affected; not sure, if command based switching includes those requests. Edit: I did a quick test and also had a short freeze on a running HD-stream. Using Sil3132 and Sil3726 PM chips. Yeah probably command based switching will have that problem. Do you know if 3132 is cmd-based or FIS-based? 3132 chipset supports FIS based switching. Those Jmicron chips found on many MoBos support command based switching. Intels latest ICHs also support commandbased switching, but at least in the win drivers usually deactivated due to unresolved bugs afaik. The combination of Sil3132 with 3726 supports FIS-based switching and is a combination that works best when using PMs, although it seems there is also freezes reported. Quote Link to comment
ftp222 Posted November 18, 2009 Share Posted November 18, 2009 I would prefer the pause, stutter or getting new hardware vs spinning up a drive I don't have to. A a tunable would be really nice. I agree with WeeboTech on this one. My whole goal with unRAID is to keep power usage and wear-and-tear on drives to a minimum. By the way, I am absolutely loving the new Samba tweaks to speed up access and directory listings. There is a huge difference from beta8 to beta11! Quote Link to comment
Kaygee Posted November 18, 2009 Share Posted November 18, 2009 Checked BIOS and sure enough my SATA controller was set to IDE. Changed it to AHCI, and now have: parity device: pci-0000:00:1f.2-scsi-5:0:0:0 (sdh) ata-Hitachi_HDT721010SLA360_STF607MH34HUTK disk1 device: pci-0000:00:1f.2-scsi-2:0:0:0 (sde) ata-ST3500630AS_5QG2X1TT disk2 device: pci-0000:00:1f.2-scsi-1:0:0:0 (sdd) ata-ST3500630AS_5QG2WW8Z disk3 device: pci-0000:00:1f.2-scsi-3:0:0:0 (sdf) ata-Hitachi_HDP725032GLA360_GEAR34RF1VTGTG disk4 device: pci-0000:00:1f.2-scsi-4:0:0:0 (sdg) ata-Hitachi_HDP725032GLA360_GEAR34RF1VSYMG disk5 device: pci-0000:03:00.0-scsi-0:0:0:0 (sda) ata-Hitachi_HDT721010SLA360_STF607MH34AXVK disk6 device: pci-0000:00:1f.2-scsi-0:0:0:0 (sdc) ata-ST3500630AS_5QG2YGPK Initial testing shows that the issue is gone. Good news. 1 down, several more to go... Quote Link to comment
WeeboTech Posted November 18, 2009 Share Posted November 18, 2009 Might some of the stuttering also be related to the BKL used by ReiserFS? Always wondered that myself. There is a significant patch in the works to eliminate the BKL in reiserfs, discussed [urlhttp://thread.gmane.org/gmane.comp.file-systems.reiserfs.general/22136]here[/url]. This patch is not in 2.6.31.x series & I haven't looked at 2.6.32 yet to see if it's in there [care to investigate this for me?]. FYI, the term "BKL" means "Big Kernel Lock". Originally with linux, when a driver or file system needed to do certain atomic operations it would acquire the BKL, do it's thing, then release the BKL. During the time BKL is held, kernel can do absolutely nothing else. There's been a longstanding effort in kernel development to get rid of the BKL because (I think) of it's impact in multi-core systems. Since the bit rates for media are relatively slow & there exists plenty of other potential bottlenecks, I don't think this is an issue whatsoever in unRAID. But interested readers can google "linux bkl" and find lots of discussion on the subject. I know this is out there a bit, but maybe there is some in house development version you have that would allow testing with EXT3. OK back to controller situation. Is there anyone with the Supermicro 8 Port controller who can post the devices and provide results. Might be good for gathering information. Quote Link to comment
BRiT Posted November 18, 2009 Share Posted November 18, 2009 Always wondered that myself. There is a significant patch in the works to eliminate the BKL in reiserfs, discussed here. This patch is not in 2.6.31.x series & I haven't looked at 2.6.32 yet to see if it's in there [care to investigate this for me?]. A quick look through the 2.6.32-rc7, it doesn't seem like the BKL is killed there. The fs/resiserfs/inode.c still issues "lock_kernel();". Quote Link to comment
Kaygee Posted November 18, 2009 Share Posted November 18, 2009 I believe Reiser 4 was supposed to remove BKL In a mailing list post from July 2009, Edward Shishkin wrote that in the coming autumn, they would start exploring the opportunity of getting Reiser4 into the main Linux kernel.[14] In a November 2009 interview[15] to Phoronix he said he is going to publish plug-in design document for independent expert review. He is currently aiming for USENIX Annual 2010. If all goes well, Reiser4 may enter mainline Linux kernel around version 2.6.36. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.