Results 1 to 14 of 14

Thread: Troubles with new 5-bay eSATA burly and Lycos PCI-X 4-port host card

  1. #1
    Join Date
    May 2008
    Posts
    4

    Default Troubles with new 5-bay eSATA burly and Lycos PCI-X 4-port host card

    We have just got a new 5-bay eSATA burly enclosure and a Lycos PCI-X 4-port host card from you guys.

    So, I installed 5 Seagate 7200.11 Barracuda drives into the enclosure and switched it on to test. Everything was OK. It turned on and the LEDs were shining. Good.

    The plan was to attach this new storage to our cluster, so I tried to test it on my desktop first. I plugged the PCI-X into a conventional PCI second slot on my Intel D945GCZ motherboard. I worked and Linux readily recognized the card and loaded the drivers. Then I tried to boot with the enclosure attached. It booted and eventually I got only one drive of 5. The one from the bottom bay. I rebooted again and went into the RAID utility, there was only one drive recognized. The number of the drive corresponded to the connector on the card. It seemed to me that the card just worked in a non-PM mode...

    Anyway, I installed the card into a real PCI-X slot on the cluster, and when I was just ready to test it. The enclosure suddenly died.... I switched it on, the LEDs went on for about 5 seconds and then all the lights switched off... This is it. It is dead at the moment.

    What do we do now? And why I saw only one drive when it worked?

    --- Dmitri

  2. #2
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,125

    Default

    If you only see one drive that usually means the Port Multiplier enclosure is attached to a non port multiplier capable host. Many internal SATA host buses are not PM capable.

    Need a lot more information. Which card? Just checking because I don't know one named Lycos, you probably meant Lycom.

    If Lycom, what driver? On what computer and what OS?

    Have you tried powering the enclosure with the drives pulled out a little. It is possible that a shorted drive can cause the power supply to power down. It has overload protection circuitry. testing would include removing all the drives and seeing if the enclosure will power fans and LEDs. Then add in one drive at a time and see if one of them is the cause of the problem.

    The more information we get the better. I personally have never used the word 'Cluster' in reference to SATA storage. So you may end up having to define where and what that means so I can be at the same place you are at.

    Rick
    molṑn labe'
    "I am a mortal enemy to arbitrary government and unlimited power. I am naturally very jealous for the rights and liberties of my country, and the least encroachment of those invaluable privileges is apt to make my blood boil."
--Ben Franklin

  3. #3
    Join Date
    Nov 2004
    Location
    Germany
    Posts
    2,352

    Default just my 2 cents

    Hello,

    more info needed.

    1. Which card, model, vendor?

    2. In which Mainboard/Computer is it installed?

    3. Which operating System is used?

    4. What driver is used (if Linux there are some non vendor drivers which could be the culprit)?

    Regards

    Nicolas
    Custom Configurations! Rad Hacks and Mods!

  4. #4
    Join Date
    May 2008
    Posts
    4

    Default details

    Quote Originally Posted by Nicolas View Post
    Hello,

    more info needed.
    Hello Guys,

    Thank you for prompt replays. OK, here are the details...

    The desktop I tried on and saw just one drive:

    Intel D945GCZ motherboard with Pentium D dual core, 2 GB RAM
    Linux Debian, kernel 2.6.18

    The enclosure:
    Burly 5 Bay w/ Port Multiplier MGBurly5PM $529.00
    http://www.macgurus.com/productpages...tMultiEncl.php

    The eSATA host (from you, as I actually mentioned): http://www.macgurus.com/productpages/sata/lycesata4.php
    Sorry, it is Lycom, I made a mistake on this.

    Seen as:

    # lspci
    0000:07:00.0 RAID bus controller: Silicon Image, Inc. SiI 3124 PCI-X Serial ATA Controller (rev 02)


    Drivers:

    # lsmod
    sata_sil24 14308 0
    libata 89332 2 sata_sil24,ahci


    The mother board has 1 PCIe and 2 PCI slots. The host card is PCI-X, it did not work in the PCI slot 1, but it did work in the PCI slot 2. It seems that slot 2 just has more features on this particular motherboard.

    However, I am pretty sure, that seeing just 1 drive did not have anything to do with Linux and drivers, because, I could enter the host card raid utility even before loading Linux (Ctrl-S) and the card itself saw only one drive of 5 (or even 2, I removed 3 drives) installed. If I remove the bottom drive, the card did not see anything at all.

    This is from dmesg after the OS was loaded:


    sata_sil24 0000:07:00.0: version 0.3
    ACPI: PCI Interrupt 0000:07:00.0[A] -> GSI 21 (level, low) -> IRQ 82
    ata5: SATA max UDMA/100 cmd 0xF89F0000 ctl 0x0 bmdma 0x0 irq 82
    ata6: SATA max UDMA/100 cmd 0xF89F2000 ctl 0x0 bmdma 0x0 irq 82
    ata7: SATA max UDMA/100 cmd 0xF89F4000 ctl 0x0 bmdma 0x0 irq 82
    ata8: SATA max UDMA/100 cmd 0xF89F6000 ctl 0x0 bmdma 0x0 irq 82
    scsi4 : sata_sil24
    ata5: SATA link down (SStatus 0 SControl 300)
    scsi5 : sata_sil24
    ts: Compaq touchscreen protocol output
    ata6: SATA link down (SStatus 0 SControl 300)
    scsi6 : sata_sil24
    ata7: SATA link down (SStatus 0 SControl 300)
    scsi7 : sata_sil24
    ata8: SATA link down (SStatus 0 SControl 300)


    Then, as I said, I tried to set it up on our cluster's master node (server): Dell PowerEdge 2850 running CentOS, kernel 2.6.9-67. There are other 65 nodes connected to the master over LAN, but this should not be important here, I guess.

    http://www.dell.com/downloads/global...2850_specs.pdf

    It has 3 PCI-X slots. So, I installed the Lycom host card into the second one, connected the server to the enclosure, both were switched off. Then, I powered on the enclosure to have it running during the server's boot. It went on for 5 seconds and then all the lights were off. At that time I had not powered up the master yet.

    The card itself seemed working OK, pretty much in the same way as on my desktop before.

    I did try to remove some drives and even all of them entirely. At the moment all the drives on my desk and the enclosure is just empty. It does not power up anyway. Just dead. Nor fans, neither lights. And I did tried different power sockets and power cords.

    The drives initially were set to 1.5 Gb/s transfer rate, but then I removed the jumpers to make them use 3.0 Gb/s rate.

    --- Dmitri.

  5. #5
    Join Date
    Nov 2004
    Location
    Germany
    Posts
    2,352

    Default

    Hello,

    on the Burly PM and the card... I don't no nothing at all
    Here you have to wait for Rick.

    Never used CentOS I tested it in the past and I didn't like it.
    I am using GenToo (Stage3build) and Fedora, also on another PC SuSe and Debian. At work mostly RHLE or Debian.

    The only thing I can tell you is there are alot of self build, modded and crapy Linux drivers out there. So if you do have the chance to test the card within a Windows PC or a Mac... try it just to proof it is NOT Linux or a messy Linux driver

    Regards

    Nicolas
    Custom Configurations! Rad Hacks and Mods!

  6. #6
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,125

    Default

    Dmitri,

    Ok, thanks for the clarity on the setup. Sounds like the power supply went poof somehow on that enclosure if you cannot get even fans and LEDs to turn on with it driveless. If nothing else we have to deal with that either by shipping you a power supply for you to swap out or by shipping the whole thing back for our guys to swap in the warehouse.

    I still don't know why you would only have the one drive bay show up when things were working. Every Burly that ships from MacGurus gets a full set of drives installed to test the enclosure before it goes out the door. We'll need to address that as well.

    I asked Brian to look into this thread and see if he can help. I am pretty much a Mac guy, he is the cross platform guy here and knows the other drivers a lot better than I do. We'll see if he has any ideas.

    When you get a chance though, send me an email at support so we can start the process on getting repairs going.

    Rick
    molṑn labe'
    "I am a mortal enemy to arbitrary government and unlimited power. I am naturally very jealous for the rights and liberties of my country, and the least encroachment of those invaluable privileges is apt to make my blood boil."
--Ben Franklin

  7. #7
    Join Date
    Nov 2004
    Posts
    66

    Default

    Hello Dmitri,

    Couple of things to check. First with the power supply issue, might pull the lid off of the enclosure to make sure that we don't have a wire pinched between the lid and the side of the case. Since the power supply worked initially, makes me wonder if we've got something shorted out inside. Generally when a PS actually dies it will hold a minimal load such as fans alone but drop off with the higher load of drives. Also check that one of the white Molex power supply connectors didn't move around in shipping and start touching a screw protruding inside the case. Any short will drop that power supply off line.

    Next issue. I don't think anybody has asked what drives you have in the enclosure. The firmware on the boards is designed to work best with standard desktop drives with the standard retail firmware. We occasionally run into problems when used with enterprise class drives. Even the custom firmware on some OEM drives can be a problem. Please let us know what you populated it with including the firmware version on the drives.

    I agree with Nicolas, it's probably a bad driver that's causing only the drive that's attached to the pass thru port on the PM board to show up. If the host doesn't have the correct PM capable driver you will only see that particular drive even if all five are attached. HERE is a link that has multiple Linux drivers, between us we should be able to find a driver that will work with CentOS. If it's not readily available we'll contact the card vendors engineer that's very helpful with driver problems.

    Let's get the power supply issue cleared up first then attack the driver problem. If you don't find anything shorted internally then it's probably quickest to get you a replacement power supply, it's easily field changeable in a few minutes.

    Brian.

  8. #8
    Join Date
    May 2008
    Posts
    4

    Default inside check

    Hi,

    Thank you again for all you help.

    I looked inside. Everything looks OK here. No shorts, nothing. I even disconnected the power supply from all the bays and even from PM card. Still dead.

    The drives, as I mentioned, are just desktop Seagate Barracuda 7200.11 500 GB, rated 0.65A +5V, 0.42 A +12 V,
    ST3500320AS
    P/N 9BX154-303
    Firmware SD15
    WWN: 5000C5000995B46C
    Made in China

    I am still having difficulty believing that the problem with just one drive being seen is in the drivers.

    As I said, at the very first stage of booting, there is a message from the host card bios is displayed on screen. This is even before the MB bios post check. That message shows the name of the card and the drives attached. There was only one drive (at best) indicated. The number of the drive corresponded to the number of card's connector used. I tried the bottom and the top connectors and got drives 0 and 3, correspondingly.

    Even more, I was able to enter the cards setup (Ctrl-S or F4 during boot) and there was only one drive. I tried to create a raid array, but the utility reported that there were too few drives connected to do that. There was no OS whatsoever loaded at that moment. How can this be a driver issue?

    And I am going to contact you at the support address too.

    Thank you again.

    Regards,

    --- Dmitri.

  9. #9
    Join Date
    Nov 2004
    Posts
    66

    Default

    In BIOS only that one drive will be seen, that's normal. The one port on the PM board is a pass through and allows the drive to be seen without the OS. I think that the reason you only see one drive with the OS is a driver problem. Since you're using drivers inherent to your version of the OS, there very well may be newer and better suited drivers available. Once we get the PS replaced we'll be able to test the enclosure with new drivers on the host card. (I think Ricks working with you on the PS replacement)

    I have no experience with CentOS and I don't believe that there are drivers listed for it specifically. There are however several Linux versions of the drivers that we can work with. That card has an extensive list of available drivers. I've already emailed their engineer to see what specific driver they recommend we start with for your OS version.

    I've also not seen that drive firmware before. It may be worth checking your drive vendor to determine if it's an OEM or retail version drive. OEM drives are probably our most common tech support issue in the last year. In many cases the firmware just isn't compatible with PM systems. I doubt that's the actual problem here but it helps to know everything about the hardware that's involved. I still think you're having a driver issue.

    Please double check on the drives just so we know what we're dealing with. I'll post or email you about drivers as soon as I hear back from Taiwan.

    Brian

  10. #10
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,125

    Default

    Dmitri,

    What voltage are you connecting to? And is the switch on the back panel set correctly for that power? Is it possible the switch is set to low voltage and your power is higher?


    Rick
    molṑn labe'
    "I am a mortal enemy to arbitrary government and unlimited power. I am naturally very jealous for the rights and liberties of my country, and the least encroachment of those invaluable privileges is apt to make my blood boil."
--Ben Franklin

  11. #11
    Join Date
    Nov 2004
    Posts
    66

    Default

    Dmitri,

    I just did a little quick reading about CentOS and PM systems. It seems that there are a lot of problems using them depending on the Kernel used. I haven't found any references to your specific kernel but will keep looking. In most cases there have been workarounds. You're using the most common PM chipsets available so that very well may help.

    Brian

  12. #12
    Join Date
    May 2008
    Posts
    4

    Default

    Quote Originally Posted by ricks View Post
    Dmitri,

    What voltage are you connecting to? And is the switch on the back panel set correctly for that power? Is it possible the switch is set to low voltage and your power is higher?


    Rick
    It is set to 115 V and here in Canada we are using 110 V, I believe. In any case it worked for a while and I did not change anything. So, I guess it is unlikely the cause of the problem.

    --- Dmitri.

  13. #13
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,125

    Default

    Thanks for the info. I didn't remember what Canadian voltage was and wanted to check. Time to swap out the power supply.

    Rick
    molṑn labe'
    "I am a mortal enemy to arbitrary government and unlimited power. I am naturally very jealous for the rights and liberties of my country, and the least encroachment of those invaluable privileges is apt to make my blood boil."
--Ben Franklin

  14. #14
    Join Date
    Nov 2004
    Posts
    66

    Default

    Dmitri,

    I've been going back and forth with the Lycom engineers trying to find a driver that will work on CentOS and provide PM capability for your cluster. Unfortunately other than confirming that it will not work in PM mode we haven't made much progress even to the point of installing version 5.1. Since most PM capable hosts use the Silicon Image chipset and the same base drivers I doubt you'll find much difference with another similar card. It may be possible to use a Marvel based card, which I'll be looking into next to verify that drivers are available. I doubt it though since the SI based cards usually have a much wider OS support than about anything else.

    This process takes time as their engineers are in Taiwan and it takes a day to get a response from them each time. Still working on it though.

    Brian.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •