bubbaQ Posted February 16, 2009 Author Share Posted February 16, 2009 My (not very limited) experience says that when software A, consistently works (along with everything else in the past) and new software B consistently doesn't, then it's software B. Not always... you can have a piece of hardware with problems or multi-vector compatibility issues that are only exposed with software B. Quote Link to comment
NLS Posted February 16, 2009 Share Posted February 16, 2009 True. Which again points us there. What is different in software B that touches the hardware. Quote Link to comment
lovingHDTV Posted February 17, 2009 Share Posted February 17, 2009 I got BubbaRaid up and running and it appears to work fine for me. I'm using the latest version. Now for a couple questions. I need to enable rsync. I use this to perform my offsite backups. How is the best way to integrate this into bubbaraid and not have it wiped out everytime I upgrade? What I have running now is: go: # Enable rsync for backups cp /boot/config/secrets /etc/rsyncd.secrets chmod 400 /etc/rsyncd.secrets rsync --daemon --config=/boot/config/rsyncd.conf rsync is already in UnRAid so all you need is the password and conf. files. Any reason why I can't just do the same thing for bubbaraid? I guess if I were really ambitious I could build a web page to support the two file, but for now I"m hoping this will work OK. thanks, dave Quote Link to comment
lovingHDTV Posted February 17, 2009 Share Posted February 17, 2009 I got BubbaRaid up and running and it appears to work fine for me. I'm using the latest version. Now for a couple questions. I need to enable rsync. I use this to perform my offsite backups. How is the best way to integrate this into bubbaraid and not have it wiped out everytime I upgrade? What I have running now is: go: # Enable rsync for backups cp /boot/config/secrets /etc/rsyncd.secrets chmod 400 /etc/rsyncd.secrets rsync --daemon --config=/boot/config/rsyncd.conf rsync is already in UnRAid so all you need is the password and conf. files. Any reason why I can't just do the same thing for bubbaraid? I guess if I were really ambitious I could build a web page to support the two file, but for now I"m hoping this will work OK. thanks, dave Houston we have a problem! It worked fine on my test box, but when I booted it up on my UnRaid box, it cannot find ide0 or id1 so I'm missing several HDD including my parity drive. When I look at the syslog I don't see ide0 or id1 even mentioned. I've not rebooted into 4.4.2 and it works fine, though parity is being rebuilt right now. Ideas? thanks, dave Quote Link to comment
RobJ Posted February 17, 2009 Share Posted February 17, 2009 True. Which again points us there. What is different in software B that touches the hardware. If BubbaQ does not mind, perhaps a fresh set of eyes might spot a clue when comparing the 2 syslogs? Do you mind making them available one more time, one with and one without BubbaRAID? Quote Link to comment
NLS Posted February 17, 2009 Share Posted February 17, 2009 of course! Here are some files (forum doesn't allow me to upload)... 1) http://www.ilogic.gr/various/log.zip is from my working system 2) http://www.ilogic.gr/various/MESSED-LOGS.zip is from my messed up system... BTW my UNRAID was enumerated as sde1, so I didn't get an sdf (should there be any device after the stick?). I have 3 IDE disks and 6 SATA all working perfectly in unRAID (also include disk.cfg in zip AND a screenshot of how unRAID "sees" those disks). 3) http://www.ilogic.gr/various/output.zip is /usr/local/sbin 4) http://www.ilogic.gr/various/user.conf.php is my wtorrent config file I didn't touch the permission of any folder... For rtorrent and wtorrent: drwxrwxrwx 7 root root 16384 Jan 1 1970 boot/ --- drwxrwxrwx 15 root root 32768 Feb 9 21:54 bubba/ --- drwxrwxrwx 2 root root 32768 Feb 9 21:38 rtorrent/ drwxrwxrwx 11 root root 32768 Feb 8 19:05 wtorrent/ (plus same permissions for everything in directory wtorrent as I already told you) For torrent destination: drwxr-xr-x 12 root root 0 Feb 10 09:15 mnt/ --- drwxr-xr-x 15 root root 352 Jan 14 22:23 disk7/ --- drwxrwxrwx 13 root root 544 Feb 12 20:34 P2P/ ...you see they are ok (/mnt and /mnt/disk7 are default as set by unRAID). ...there you are... thanks! Quote Link to comment
bubbaQ Posted February 18, 2009 Author Share Posted February 18, 2009 NLS, you have multiple problems. Please give up on the wtorrent issue. If the wTorrent people don't see fit to fix it, there isn't anything I can do. Repeatedly bringing up the wTorrent issue just clouds the water. I compared your syslogs some time ago. Your hardware problem is with the USB subsystem. Something is not being properly initialized, and appears to be hanging long enough that it is not ready when the call to mount the flash by volume name is made, but it then does finish initing later, so you can mount the flash by hand. But by that time, too many other things have failed due to the faile mount of the flash as listed in fstab. Quote Link to comment
RobJ Posted February 18, 2009 Share Posted February 18, 2009 I can't help with the torrent stuff at all. And if there is any confusion between my analysis and that of BubbaQ, then I defer to his experience. I would however like to present my own analysis of what I saw in the comparison of syslogs. I don't have good news, as I can see the serious event that precipitated all of the subsequent failures, and I can roughly explain why most of the drives and networking are either not visible or not working, but I can't explain why it happened or what to do to fix it, although I do have a possible theory below. The problem is that an interrupt was raised on IRQ 18, and nobody claimed it ("kernel: irq 18: nobody cared"), so after its attempt to debug it and display a short execution trace, it then disables IRQ 18 ("kernel: Disabling IRQ #18"), which is disastrous. IRQ 18 had been assigned to that entire controller, so none of the 3 IDE drives connected are able to finish their setup, and a long series of exceptions, timeouts, and resets begin, which appear to hang the machine too long for anything else to initialize. IRQ 18 was also assigned to the networking, so that would not have worked either, if it had been able to progress that far. IRQ 18 was also one of the IRQ's assigned to *some* of the USB support, but not apparently to the ports used by your flash drive or the USB keyboard and mouse. So the flash drive was setup fine as sde, but the rest of the 'life' of this computer was just involved in the setup of the 3 IDE drives (sdf, sdg, and sdh), and after many minutes, it *did* get them partly setup and assigned, but not usable, and too late for anything else. There is one more thing that looks wrong to me, and that leads me to a wild theory, but I could be far off the mark. Both syslogs have a very similar setup until after the Promise card, which is the first drive controller they come to. At this point, they begin to work on the IDE controller with 3 IDE drives attached. The stock unRAID kernel uses the piix driver to handle it, without issue. The BubbaRAID kernel uses ata_piix instead, and gets as far as identifying the 2 IDE drives on the primary channel and configuring them for UDMA/100, when the 'unclaimed' IRQ 18 occurs, and almost everything goes wrong from that point on. It never gets to the last drive controller, with SATA drives in an IDE emulation mode that do need the ata_piix driver. I'm not an expert in this, but in my experience the piix driver is supposed to handle true IDE drives, and the ata_piix driver handles SATA drives pretending to be IDE drives, but not true IDE drives. That is, it handles SATA drives on a motherboard with BIOS settings in IDE emulation mode. So it looks to me as if the stock kernel is configured correctly, to assign piix to these IDE drives, and the BubbaRAID kernel for unknown reasons assigns the wrong driver ata_piix, which perhaps is what causes their drive support to collapse. My wild theory? SATA drives and IDE drives talk a different language, a different protocol. (That's why you can't add NCQ support to PATA drives.) So I wonder if the first IDE drive sdf made its first communication, by raising IRQ 18, and ata_piix in effect said "that's IDE talk, not SATA talk, so that's not for me", and passed on it, leaving the IRQ unhandled. As to why NLS is the only one affected, this is just my own idea so could be wrong, but I can't help thinking that users that are more technically advanced to use BubbaRAID, seem more likely to me to have more modern equipment, less likely to try to use this with older IDE drives that need the piix driver. Just a guess, with very little data behind it... Quote Link to comment
bubbaQ Posted February 18, 2009 Author Share Posted February 18, 2009 Thanks Joe. I saw the piix issue, but it doesn't seem to me to be the *cause* but just a symptom. I just grepped for piix in both the BubbaRaid .config and the stock unRAID .config. The stock unRAID .config, has the following: =============================== CONFIG_BLK_DEV_PIIX=m CONFIG_ATA_PIIX=m # CONFIG_PATA_MPIIX is not set # CONFIG_PATA_OLDPIIX is not set The BubbaRaid .config uses: ============================ CONFIG_BLK_DEV_PIIX=m CONFIG_ATA_PIIX=m # CONFIG_PATA_MPIIX is not set # CONFIG_PATA_OLDPIIX is not set My current theory is something on the SMBus.... but it is hard to debug that w/o the machine in front of me. BubbaRaid enables sensors on the SMBus for hw monitoring, but those are all enabled in 0.18 (which works for NLS) and nothing there has changed. I'm building a new kernel for NLS tonight, and will see what happens. Quote Link to comment
prostuff1 Posted February 18, 2009 Share Posted February 18, 2009 A lot of technical stuff RobJ and bubbaQ, you guys are brilliant. RobJ's explanation pretty much made sense to me, and while i do not consider myself dumb it was very well explained. I would never have been able to deduce that stuff form his syslogs (and i even downloaded them to take a look). The only other thing i have a question about is this. I have been looking around at the config files in BubbaRaid and I see that most if not all of the stuff required to get this running is already there. I am not sure how to proceed any further then that, without possibly screwing something up. I have no idea how tightly wtorrent is tied to rtorrent and how exactly they communicate. For now i have reverted to .21 and wtorrent is working again. I will probably try to upgrade to the latest version again this weekend and see how it goes. I am running a set of preclears on some new data disks i am adding to the array, so i don't/can't take it offline. Quote Link to comment
lovingHDTV Posted February 18, 2009 Share Posted February 18, 2009 So could the reason NLS is not getting his IDE drives initialized be related to my problem of not getting the ide drives initialize? I don't see time outs or anything, but I do not ever get the ide0 nor ide1 controllers discovered with bubbaraid. 2 and 3 come in fine. . . My hardware is the orignal MB, promise ide cards. thanks, dave Quote Link to comment
bubbaQ Posted February 18, 2009 Author Share Posted February 18, 2009 I'm also going to do a compare of the modules from stock to Bubba, and see if something is being built differently (other than version magic). But as Joe pointed out, the difference is at the Promise ctrlr load. But I both ata_piix and piix are needed for a system with both SATA and ATA drives -- so I can't just remove the ata_piix module. Plus, they both exist as modules in stock, so they obviously can live together. Something *else* is causing the Promise card to be handled by ata_piix instead of piix. That's when it starts cascading. Quote Link to comment
NLS Posted February 18, 2009 Share Posted February 18, 2009 I'll get back to you tonight (Greek TZ) Quote Link to comment
lovingHDTV Posted February 18, 2009 Share Posted February 18, 2009 I just compared my stock unraid 4.4.2 syslog with bubbaraid syslog. I see them diverge at line 400 when in stock unraid it finds ide0 and ide1, with the piix driver, while bubbaraid uses the ata_piix driver and calls them ata1 and ata2. These are the two ide interfaces on the motherboard. bubbaraid: Feb 18 13:29:04 Tower kernel: ata_piix 0000:00:1f.1: version 2.12 Feb 18 13:29:04 Tower kernel: ata_piix 0000:00:1f.1: enabling device (0005 -> 0007) Feb 18 13:29:04 Tower kernel: ata_piix 0000:00:1f.1: PCI INT A -> GSI 18 (level, low) -> IRQ 18 Feb 18 13:29:04 Tower kernel: ata_piix 0000:00:1f.1: setting latency timer to 64 Feb 18 13:29:04 Tower kernel: scsi1 : ata_piix Feb 18 13:29:04 Tower kernel: scsi2 : ata_piix Feb 18 13:29:04 Tower kernel: ata1: PATA max UDMA/100 cmd 0x1f0 ctl 0x3f6 bmdma 0xffa0 irq 14 Feb 18 13:29:04 Tower kernel: ata2: PATA max UDMA/100 cmd 0x170 ctl 0x376 bmdma 0xffa8 irq 15 Feb 18 13:29:04 Tower kernel: ata1.00: ATA-6: WDC WD3200JB-22KFA0, 08.05J08, max UDMA/100 Feb 18 13:29:04 Tower kernel: ata1.00: 625142448 sectors, multi 16: LBA48 Feb 18 13:29:04 Tower kernel: ata1.00: configured for UDMA/100 Feb 18 13:29:04 Tower kernel: ata2.00: ATA-6: ST3120026A, 8.01, max UDMA/100 Feb 18 13:29:04 Tower kernel: ata2.00: 234441648 sectors, multi 16: LBA48 Feb 18 13:29:04 Tower kernel: ata2.01: ATA-7: ST3250620A, 3.AAF, max UDMA/100 Feb 18 13:29:04 Tower kernel: ata2.01: 488397168 sectors, multi 16: LBA48 Feb 18 13:29:04 Tower kernel: ata2.00: configured for UDMA/100 Feb 18 13:29:04 Tower kernel: ata2.01: configured for UDMA/100 Feb 18 13:29:04 Tower kernel: scsi 1:0:0:0: Direct-Access ATA WDC WD3200JB-22K 08.0 PQ: 0 ANSI: 5 stock UnRaid: Feb 17 12:19:30 Tower kernel: piix 0000:00:1f.1: IDE controller (0x8086:0x24db rev 0x02) Feb 17 12:19:30 Tower kernel: PIIX_IDE 0000:00:1f.1: enabling device (0005 -> 0007) Feb 17 12:19:30 Tower kernel: PIIX_IDE 0000:00:1f.1: PCI INT A -> GSI 18 (level, low) -> IRQ 18 Feb 17 12:19:30 Tower kernel: piix 0000:00:1f.1: not 100%% native mode: will probe irqs later Feb 17 12:19:30 Tower kernel: ide0: BM-DMA at 0xffa0-0xffa7 Feb 17 12:19:30 Tower kernel: ide1: BM-DMA at 0xffa8-0xffaf Feb 17 12:19:30 Tower kernel: Probing IDE interface ide0... Feb 17 12:19:30 Tower kernel: hda: WDC WD3200JB-22KFA0, ATA DISK drive Feb 17 12:19:30 Tower kernel: hda: host max PIO4 wanted PIO255(auto-tune) selected PIO4 Feb 17 12:19:30 Tower kernel: hda: UDMA/100 mode selected Feb 17 12:19:30 Tower kernel: Probing IDE interface ide1... Feb 17 12:19:30 Tower kernel: hdc: ST3120026A, ATA DISK drive Feb 17 12:19:30 Tower kernel: hdd: ST3250620A, ATA DISK drive Feb 17 12:19:30 Tower kernel: hdc: host max PIO4 wanted PIO255(auto-tune) selected PIO4 Feb 17 12:19:30 Tower kernel: hdc: UDMA/100 mode selected Feb 17 12:19:30 Tower kernel: hdd: host max PIO4 wanted PIO255(auto-tune) selected PIO4 Feb 17 12:19:30 Tower kernel: hdd: UDMA/100 mode selected Feb 17 12:19:30 Tower kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14 Feb 17 12:19:30 Tower kernel: ide1 at 0x170-0x177,0x376 on irq 15 After this it continues booting and the array does not start because it is missing three drives. Hope this helps a little, dave Quote Link to comment
foo_fighter Posted February 18, 2009 Share Posted February 18, 2009 First a big thanks to BubbaQ and the rest of the people on this thread/forum. I thought I would post some of my experiences installing and configuring BubbaRaid since it was a little hard to find out how to get a few things working. I already had a basic version of unraid running, so adding bubbaraid by unzipping and running the enable_bubba.bat in a windows shell was pretty straight forward. It booted up immediately and I could access the portal. First slimserver: installing via the provided link was also pretty painless. Is that bug in 7.3.1 fixed yet? Has anyone tried the latest versions(7.4..)? SqueezeCenter is at http://192.168.0.100:9000/ (substitute your UnRaid's IP) SoftSqueeze is at: http://192.168.0.100:9000/html/softsqueeze/webstart.html The first problem was that Squeezecenter could not see all of my directories and files. I need to chmod all the directories to world read+executable (chmod 755) and all the files to world readable (chmod 644) To do this you can use something like this after logging into your box: #change all directories and subdirectories in current path: find . -type d -exec chmod 755 {} \; #change all files in current and sub directories find . -type f -exec chmod 644 {} \; -exec works better than | xargs if you have spaces in file or directory names To set newly created files, you have to modify the samba configuration by using the go script: For some reason this didn't work in the go script(/boot/configs is where I keep my custom configs): echo "Setting Share Permissions" cp /boot/configs/smb.shares /etc/samba smbcontrol smbd reload-config But this did(got it from another thread): #Try reseting samba again #define USER_SCRIPT_LABEL Set SAMBA File Permissions #define USER_SCRIPT_DESCR Sets SAMBA file permissions to 0644 and directory permissions to 0755 echo "<pre>" set -xv sed -i -e "s/create mask = 0711/create mask = 0644/" /etc/samba/smb.shares sed -i -e "s/directory mask = 0711/directory mask = 0755/" /etc/samba/smb.shares smbcontrol smbd reload-config As for SoftSqueeze, I had a problem where my browser's connection was incorrectly set to look for a proxy or auto detect. This resulted in a security exception unknown protocol socket message from java. To fix this I set the connection to no-proxy in mozilla since by default java looks at the default browser's config, you can also do this in the java control panel. For some custom bash environment settings and aliases, I added to the go script: echo "Copying bashrc files" cp /boot/configs/.bash_profile /root For the Powerdown script: (downloaded from http://code.google.com/p/unraid-powercontrol/ ) echo "Installing Powerdown packages" CTRLALTDEL=YES SYSLOG=YES STATUS=YES START=YES installpkg /boot/packages/powerdown-1.02-noarch-unRAID.tgz For lm_sensors: Run sensors-detect to detect any supported chips, then I added: echo "Starting lm_sensors" cp /boot/configs/lm_sensors /etc/sysconfig modprobe w83627ehf sensors -s Actually I'm not sure if I need the sysconfig line. For apcupsd: I have a cyberpowersystems UPS, but I only had to modify these lines with the web interface to the apcupsd configuration file: UPSNAME CyberPower UPSCABLE usb UPSTYPE DEVICE Now for some questions: I tried running s2ram -n and it could not detect my system, but s2ram -f seemed to work. It would go into low power and I could WOL. Now how can I configure it to power down after a set amount of time? Can I add a configuration setting/button to the Portal to set this? Maybe it can look at the disk spindown time and be X minutes after the last disk spins down. What kind of housekeeping needs to be done before and after calling s2ram? I'm assuming it'd be prudent to close all the files and maybe stop a few things then resume them after waking. I don't know what would happen if the system was in s3 and a power outage happened, would apcupsd still shut it down cleanly? Thanks again! Quote Link to comment
bubbaQ Posted February 18, 2009 Author Share Posted February 18, 2009 The first problem was that Squeezecenter could not see all of my directories and files. Yup. BubbaRaid has no idea where your media files are.... and SS needs the permissions on those files to be world readable. Has anyone tried the latest versions(7.4..)? It's broke. Don't use it. I don't know what would happen if the system was in s3 and a power outage happened If you don't have a UPS, the system loses power and will reboot when powered back up. would apcupsd still shut it down cleanly? Nope. I'm not intending to hook s2ram into the BubbaRaid interface.... too much danger. I added so people can experiment with it. You'll have to handle the housekeeping and the timing yourself. I would spin up all drives and sync at a MINIMUM. Also, in S3 sleep mode, the system only pulls a couple of Watts, so even a small a UPS can last for many hours. Quote Link to comment
NLS Posted February 18, 2009 Share Posted February 18, 2009 ...btw I tried your latest internal beta and is no go for me again. I think RobJ and lovingHDTV may have a point since it's clear to me that something doesn't go ok with the drives. Not ONE drive (where it could be the USB stick) but more than one drives from "ATA4" and on... I remind again: My setup is not TOO old (even though using 3 IDE disks because I happened to have them), uses only stock mobo (no extra cards) which is a very eponymous (and once very high-end) Asus P4C800-E Deluxe with AFAIK the latest available firmware. Note also that no version before 0.01 had problems (and now that I reverted to .21 again, it works fine), all unRAID (well not all... I had to use irqpoll a couple of versions back) and Windows (XP, Server 2003 and Vista) were working fine on it (until it got "support" duty with unRAID). Quote Link to comment
bubbaQ Posted February 18, 2009 Author Share Posted February 18, 2009 2 of my 3 test machines are ASUS... K8N-E and P5B-Deluxe. I also am using the mobo controllers and no additional cards on those mobos. Quote Link to comment
NLS Posted February 18, 2009 Share Posted February 18, 2009 So what do you suggest? Quote Link to comment
lovingHDTV Posted February 18, 2009 Share Posted February 18, 2009 Hmm I'm using the original equipment. Anyone out there that using the original equipment have bubbaraid running? dave Quote Link to comment
foo_fighter Posted February 19, 2009 Share Posted February 19, 2009 I would spin up all drives and sync at a MINIMUM. Also, in S3 sleep mode, the system only pulls a couple of Watts, so even a small a UPS can last for many hours. Would it be possible to post a short shell script that does this? In fact how can I find out the command line equivalents of stop / start / spin up / spin down buttons on the Unraid main page? Thanks. Quote Link to comment
jimwhite Posted February 19, 2009 Share Posted February 19, 2009 NLS and BUbbaQ.... your Asus boards are running different Intel controller chips... ICH8 and ICH9, I believe... significant? Quote Link to comment
bubbaQ Posted February 19, 2009 Author Share Posted February 19, 2009 Yes, but the ICH driver is the same. Quote Link to comment
oconnellc Posted February 22, 2009 Share Posted February 22, 2009 First, bubbaq, thanks a ton for all the hard work you have put into bubbaraid! Second, now that I have greased the wheels... a couple questions. First, I downloaded and installed the latest (0.01.02-Beta) and ran the script to enable bubbariad. Rebooted everything and everything came up as expected. Very happy... Now, I try to configure unraid_notify and have it send a test email and I get errors: /usr/bin/unraid_notify: line 327: [: 27: unary operator expected /usr/bin/unraid_notify: line 327: [: 26: unary operator expected /usr/bin/unraid_notify: line 388: [: 481: unary operator expected /usr/bin/unraid_notify: line 396: [: 1: unary operator expected So, I figure I will telnet into the box and look at the unraid_notify script. Well, short story, I can't because my go script has been modified and it isn't installing my packages on startup anymore. So, I was able to add the appropriate lines back into the bubbaraid go script and reboot and all my packages get installed. So, is this the right way to control mods to the go script? Should I track everything that I do so that if I make a mod into one version, then I will have to reimplement those changes into the other version (i.e. with or without bubbaraid)? What about upgrades to bubbaraid? Will upgrading cause me to have to make those changes again? Next, what about those errors I am getting? I looked at the offending lines and they seem pretty harmless. I'm not sure how they could be causing a problem (although, I have to admit, I don't even know what language that script is written in, much less know for sure if the syntax is correct). Here is line 396: if [ $iMinSinceLastSend -ge $iNotifyDelay ] I'm guessing that there is a problem somewhere else manifesting itself at the reported line numbers. Any ideas anyone? Thanks in advance for any help, Chris Quote Link to comment
oconnellc Posted February 22, 2009 Share Posted February 22, 2009 Also, another question about configuring the 'unraid_notify' script. What are appropriate values for the wait between status checks and the spindown time. I'm not sure what the mechanism for checking the drives and then spinning them down is, so I don't want to use a setting that causes this checking mechanism to keep spinning up the drives to test them and then not allow then to spin down. I'm sure that there probably isn't any 'right' answer, but there may be 'wrong' answers, and I don't want to ignorantly set them to the wrong values. Thanks, Chris 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.